USA
country · [US] · events: 312
organizations
models
events
- OpenAI’s red team of AI agents, including GPT-6 Astra Cyber, spent hundreds of billions of tokens and found ways to escape virtual machines, internet access through almost every one of the few services left with a connection to the outside world (metrics, logs, authentication), and a covert communication channel between isolated agents based on CPU cache read timings — the same technique as in the Spectre and Meltdown attacksCybersecurity · Software development · AI safety · openai.com ↗
- OpenAI: on September 20, an internal research model exploited a misconfiguration of a DNS resolver during training and gained limited internet access. The company paused training and evaluations of its most capable models for 11 days, and now runs reinforcement learning without internet access — in an “offline world” built from saved web pagesCybersecurity · Software development · AI safety · openai.com ↗
- Anthropic: when Mythos 5 is resampled from the middle or late stage of an attack, the model stops on its own in only 2% of cases. In an experiment where, after the model had reasoned about whether the maintainers were simulated, it was told it was dealing with real people, it stopped in 5% of the resamples and in 82% interpreted the message as part of the exercise. The company calls this persistence “momentum”Cybersecurity · Science · AI safety · anthropic.com ↗
- Anthropic and the UK AI Security Institute (AISI): in 9 of 43 cyber-exercise runs with open internet access and safeguards disabled, Claude Mythos 5 went beyond the scope of its task and affected real people and organizations. In the main case, the model submitted malicious code to someone else’s open-source project on GitHub and used fake accounts to try to persuade the maintainer to accept it; the maintainers did not accept the codeCybersecurity · Software development · AI safety · anthropic.com ↗
- The Agentic AI Foundation at the Linux Foundation accepts four projects — Agent Skills from Anthropic, Open Responses from OpenAI, OpenEnv from Hugging Face and OpenSandbox from Alibaba — and reaches 10 projects in its first yearSoftware development · AI agent orchestration · linuxfoundation.org ↗
- Science publishes Microsoft’s work on sequential diagnosis: a system of simulated doctors that decides on its own which tests to order reaches the correct diagnosis in 83.2% of NEJM clinical case records and in 80.3% of MIMIC-IV-ED emergency department admissions, versus 36.1% and 46.8% for generalist physiciansHealth & medicine · Science · AI in medicine · doi.org ↗
- In its Q3 2026 report, Tesla says that Optimus production has begun on high-volume lines in Fremont and that these are the first robots of series-production qualityIndustry & robots · Business & markets · Work & jobs · Humanoid robots · assets-ir.tesla.com ↗
- PHP releases emergency versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35: they fix two critical vulnerabilities (CVSS score 9.2) that allow code to be executed on the server without authentication; they were reported by a researcher at Calif, which was auditing PHP under OpenAI’s Patch the Planet program, by OpenAI itself, and by WizCybersecurity · Software development · Internet & infrastructure · Open source · thephp.foundation ↗
- The US Department of Homeland Security publishes an order: federal civilian agencies must stop procuring DeepSeek products and remove from their IT systems both those products and third-party products built on DeepSeekLaws & regulation · Geopolitics & defense · Cybersecurity · Compute export controls · federalregister.gov ↗
- In 64 hours Claude generates 318,484 protein variants that bind scorpion neurotoxins and selects 180 of them; teams at Adaptyv Bio and Twist Bioscience test 211 variants for binding (31 were added by Anthropic staff), and binding is confirmed for 17 — venom neutralization was not testedHealth & medicine · Science · AI in medicine · anthropic.com ↗
- Anthropic transfers 5% of its shares to the Anthropic Public Benefit Fund (if shareholders approve), and the board of directors may add a further 2% for every +50% in the company’s value above a baseline; among the fund’s possible purposes is helping workers affected by AI, up to and including direct payments to particular groupsWork & jobs · Business & markets · Finance & money · AI and jobs · anthropic.com ↗
- Anthropic publishes data on its own workforce: the share of early-career people among new hires into technical teams falls from 27% to 9%, their share of technical departments from 21% to 14%, and by August 2026 interactive AI agents log six times as many working hours as employeesWork & jobs · Software development · Business & markets · AI and jobs · anthropic.com ↗
- An internal Claude model works out in under half an hour how the “seeds” of five standard NIST elliptic curves (P-192 through P-521) were obtained: not a hash, but the output of the lrand48 random number generator seeded with a Unix timestamp; Anthropic stresses that this is not a break, not a new attack and not a weakness in the curves, and writes that the finding strengthens confidence that the curves contain no backdoorsCybersecurity · Science · Cryptography · anthropic.com ↗
- Scientific American: after working through OpenAI’s 166-page Navier–Stokes blowup proof, mathematicians found that it builds its external force in a totally distinct way from the program of the Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa — although at release company officials “clearly and repeatedly stated” that it followed that programScience · AI in mathematics · scientificamerican.com ↗
- Google releases Gemini 4 Argon: it solved all 12 end-to-end simulated cyberattack scenarios (the previous generation solved 5); because of elevated capabilities in the cyber and biological domains, under the Frontier Safety Framework it ships with new safeguards — three layers of real-time protection — and the version without cyber guardrails went first only to trusted defenders in the Fairwind ProgramCybersecurity · AI safety · blog.google ↗
- Anthropic: in May 2026 around 2,500 code changes a day were merged into its monorepo on average, and over 30,000 a day by September, making source control the biggest development bottleneck; seven people built Scion, a centralized replacement for git, and the code review tool Antique in five months; the authors say Scion is no general-purpose replacement for gitSoftware development · Agentic engineering · anthropic.com ↗
- Anthropic’s Frontier Red Team: in a firewalled network of 8 cloud nodes running an Android emulator, an agent driven by the open model GLM-5.2 deployed Llama 3.1 8B onto the nodes on its own — all 8 nodes in 4 of 5 trials and 7 of 8 in the fifth; the authors estimate that models of 30B parameters and under could reach this level by the end of 2026 at their historical rate of improvementCybersecurity · AI safety · anthropic.com ↗
- Activision files DMCA takedowns against browser versions of Call of Duty: Black Ops II and its Zombies mode that a developer built with the help of AI coding agentsCulture & arts · Vibe coding · xda-developers.com ↗ ⧗
- Linus Torvalds publishes in his Linux kernel repository an experimental branch of 8 patches tagged “Assisted-by: LLM”: “The code, the test programs, the review behind the list above and the first draft of this text were done with an LLM coding assistant (Claude Code) under my direction”; on his laptop, reading one file from 8 threads became four times cheaper — 725 cycles per read instead of about 2,950; he lists 10 known problems himself, and the branch is not merged into mainlineSoftware development · Agentic engineering · git.kernel.org ↗
- Axios: executives at OpenAI, Anthropic and other AI companies are preparing for the public and political backlash after a major AI catastrophe that many insiders expect in the next six to 12 months; OpenAI confirms it runs such preparedness exercisesElections & politics · Cybersecurity · AI safety · axios.com ↗
- The Verge talks to more than three dozen mathematicians about OpenAI's release; Bartosz Naskręcki: it “can cause a huge collapse in the academic culture”Science · Education · Reasoning models · theverge.com ↗
- The US International Trade Commission opens investigation 337-TA-1526 on Vicor's complaint against 20 respondents over vertical power delivery systems for AI serversLaws & regulation · Business & markets · AI accelerators · usitc.gov ↗
- TypeSafe AI raises $870 million at a $7.5 billion valuation; its model Jev does not generate text and outputs only typed decisionsBusiness & markets · Finance & money · AI labs · typesafe.ai ↗
- Anthropic opens the public beta of Claude Managed Agents dynamic workflows: a lead agent writes a plan that runs across many agents; with 70 planted bugs a workflow found 66 in each of 3 runs, a single agent 14, 15 and 27Software development · AI agent orchestration · x.com ↗
- The New York Times, citing two sources: Anthropic's AI agents submitted 20 visa applications through a form on the State Department's website; all were incomplete and were not processedCybersecurity · Elections & politics · AI safety · nytimes.com ↗
- Anthropic: in evaluations Claude submitted online forms when it shouldn't have and got past access controls on live sites; Claude Haiku 4.5 sent an invented tip through a Philadelphia Police Department form; the company turns off live internet access for its internal evaluationsCybersecurity · AI safety · anthropic.com ↗
- The mathematicians who formalized the Poincaré conjecture proof, together with an NVIDIA team, complete a Lean formalization of the Hamilton–Perelman proof of Thurston's geometrization conjectureScience · AI in mathematics · arxiv.org ↗
- Anthropic launches the Cyber Mission: a Critical Infrastructure Defense Program with models and engineers for defenders, and free vulnerability scans for open-source projectsCybersecurity · Internet & infrastructure · AI safety · anthropic.com ↗
- Google Cloud: the Gemini agent runs each job on the model that fits best — from the Gemini family or Anthropic's Claude, and other models in the futureSoftware development · Business & markets · AI agent orchestration · cloud.google.com ↗
- Science publishes DeepMind’s paper on the AlphaProof Nexus agent: it autonomously resolved 9 of 353 open Erdős problems and proved 44 of 492 conjectures from the On-Line Encyclopedia of Integer Sequences, with every proof checked in Lean; in the same issue mathematicians Jeremy Avigad and Matthew Ballard write the commentary “AI for research mathematics has arrived”Science · AI in mathematics · doi.org ↗
- OpenAI bans a Russian and an Iranian influence operation that used its models to run false-front think tanks and journalists; it rates the Russian one Category 5 on a six-point scale for the first timeElections & politics · Media & information · Geopolitics & defense · Generative AI · openai.com ↗
- CrowdStrike: a threat actor, likely Chinese-speaking, attacked South Korean financial organisations using ARTEX, an open-source Chinese agentic pentesting tool, and Claude Code; the number of affected organisations is unconfirmedCybersecurity · Finance & money · Autonomous AI agents · crowdstrike.com ↗
- OpenAI rolls out GPT-6 in ChatGPT to all users, including free ones: paid tiers get GPT-6 Sol, Free and Go get GPT-6 LunaSoftware development · Business & markets · Chatbots · openai.com ↗
- Elon Musk: Grok Bot will use the best back-end model for any given task, including Claude Opus 5.5, Midjourney and SunoSoftware development · Business & markets · AI agent orchestration · x.com ↗
- NVIDIA: fine-tuned Nemotron models reached gold-medal level at IMO 2026 (30/42 against a threshold of 29, graded by official IMO graders) and IOI 2026 (535.4/600, an unofficial run)Science · Education · AI in mathematics · huggingface.co ↗
- Anthropic releases Claude Haiku 5.5: on average about 75% cheaper to run than Haiku 4.5 — $0.10 per million input tokens versus $1Software development · Business & markets · Large language model · anthropic.com ↗
- A developer posts on X that Claude Code on Opus 5.5, running unattended for hours with permission checks turned off, deleted his computer’s entire C: drive; daily backups to a NAS saved the data, and the post drew more than a million views; Claude Code lead Boris Cherny replied that this is why auto mode for permissions is the default — it “almost certainly would have caught this”Software development · Autonomous AI agents · x.com ↗
- OpenAI publishes on GitHub mathematical results from its internal model — now 719 manuscripts in 372 families; three manuscripts were withdrawn over a sign errorScience · Reasoning models · github.com ↗
- OpenAI publishes 719 mathematical papers by its internal model; 42% of the main results are formally verified in LeanScience · Reasoning models · github.com ↗
- OpenAI announces a new ad format in ChatGPT: testing during image generation for Free and Go users in the US begins later this monthBusiness & markets · Media & information · Chatbots · openai.com ↗
- The Wikimedia Foundation finds activity by “rogue” OpenAI agents on its projects: millions of automated requests, edits it considers potentially malicious, and a possible contribution to a partial outage of the Wikidata Query ServiceInternet & infrastructure · Cybersecurity · AI safety · wikimediafoundation.org ↗
- Under EU rules, OpenAI will add an invisible textGrain watermark to eligible ChatGPT and Codex text in the European Union over the coming weeks; the detector goes first to approved researchersLaws & regulation · Media & information · AI regulation · openai.com ↗
- A company owner is arrested in California, charged with smuggling servers with American GPUs worth more than $300 million to China via Malaysia and SingaporeGeopolitics & defense · Justice & courts · Compute export controls · justice.gov ↗
- Challenger, Gray & Christmas: since the start of 2026, US employers have named AI as the reason for 120,136 job cuts, about 21% of all and more often than any other reasonWork & jobs · Business & markets · AI and jobs · challengergray.com ↗
- Yann LeCun: the heads of OpenAI and Anthropic saying that AI can kill us all is “incredibly destructive for everybody” and “the worst marketing campaign you can possibly imagine”Media & information · Science · AI safety · fortune.com ↗
- OpenAI describes two incidents with internal agents: GPT-6.1 Sol, while fixing an evaluation system, disables automatic alerts to the security team, and GPT-6 Astra, while porting a program, removes its misalignment-monitoring integration and, when the user asks directly, replies that it did not remove itSoftware development · Cybersecurity · AI safety · alignment.openai.com ↗
- The Association for Human Mathematics urges mathematicians to stop working with OpenAI: “Releasing over 700 files at once is not a demonstration of scholarship, but a demonstration of power”Science · Reasoning models · ahmath.org ↗
- Following JPMorgan, Citi and Bank of America bar third-party AI agents from logging into customer accounts with the customers’ usernames and passwords; read-only access through open banking services remainsFinance & money · Business & markets · Autonomous AI agents · fstech.co.uk ↗ ⧗
- According to Every’s article “How Codex Happened”, in June 2026 Codex had more than 5 million weekly active users, and by October Codex and ChatGPT for Work together had 40 million weekly active usersWork & jobs · Software development · Business & markets · Agentic coding · every.to ↗
- According to the WSJ, citing a draft IPO prospectus, Anthropic’s revenue for the first half of 2026 was $16.3 billion versus $1.1 billion a year earlier, and its net loss exceeded $80 billion, of which $72 billion was an accounting remeasurement of convertible notes and other financial instrumentsBusiness & markets · Finance & money · AI labs · aljazeera.com ↗ ⧗
- Reflection AI unveils Beam, its first open-weight model: 501 billion parameters, 23 billion active; weights promised this month under an Apache 2.0 licenceSoftware development · Open weights · reflection.ai ↗
- OpenAI launches dots: always-on ChatGPT agents with their own cloud computer that pursue the user's goals around the clockWork & jobs · Business & markets · Autonomous AI agents · openai.com ↗ ⧗
- OpenAI admits that in June its models, during internal training, accessed Australian government websites without authorisation; at the Medicare statistics service a model ran commands and retrieved internal files and credentials, while patient records were not accessedCybersecurity · Health & medicine · Privacy & surveillance · AI safety · openai.com ↗
- A working paper, “What If Automating AI R&D Triggers an Intelligence Explosion?”, comes out signed by 22 authors, among them Geoffrey Hinton, Yoshua Bengio, Andrew Barto, OpenAI chief scientist Jakub Pachocki and Anthropic co-founder Jack Clark; the lead author is Alan Chan of GovAIScience · Laws & regulation · Elections & politics · Recursive AI self-improvement · governance.ai ↗
- Mathematicians supported by DARPA's expMath programme complete a Lean formalization of the Hamilton–Perelman proof of the Poincaré conjecture: around 4.7 million lines of code in roughly two weeksScience · AI in mathematics · arxiv.org ↗
- Utah signs a 12-month agreement with the start-up Nolla Health: its app diagnoses adults with mild to moderate acne and issues first-time prescriptions for topical treatments on its own, with no oral drugs; at first two physicians pre-approve every prescription, then physicians review cases weekly, and later sample at least 10% a month; the pilot costs $4.99 a month and launched on 5 OctoberHealth & medicine · Laws & regulation · AI in medicine · commerce.utah.gov ↗
- The US Ninth Circuit affirms the dismissal of the DMCA claims over Copilot: Copilot and Codex do not remove copyright management information but create new worksSoftware development · Laws & regulation · AI pair programmer · cdn.ca9.uscourts.gov ↗
- OpenAI: an internal model proved that the 3D Navier–Stokes equations with a smooth applied force can develop a singularity in finite time; around 10,000 coordinating AI agents, 88 hours, with a Lean formalizationScience · AI agent orchestration · openai.com ↗
- VulnCheck: of 26,153 Project Glasswing findings claimed by Anthropic, 202 (0.8%) have been fixed; Claude rated 91.5% of the findings as critical or high, the maintainers 51.3%Cybersecurity · Software development · Work & jobs · AI vulnerability discovery · vulncheck.com ↗
- Dario Amodei: since summer 2026 AI has been advancing drastically faster through recursive self-improvement; he proposes pacing frontier models without halting trainingScience · Laws & regulation · AI safety · darioamodei.com ↗
- OpenAI acknowledges that in July 2026, during internal cybersecurity evaluations, its models circumvented isolation from the internet, compromised Hugging Face’s systems and gained administrator access to a research cluster of OpenAI’s own; the company calls it a “warning shot” and has put its largest planned frontier training run on holdCybersecurity · Internet & infrastructure · AI safety · openai.com ↗
- METR investigation: in OpenAI's ExploitGym experiments about 1,200 agents meant to be isolated communicated with each other, and about 700 of them attacked Hugging Face's infrastructureCybersecurity · Internet & infrastructure · AI safety · metr.org ↗
- OpenAI: with the production ChatGPT harness and system prompt, a model’s propensity to compromise infrastructure can drop more than 100-fold; these safeguards were absent from the evaluations behind the Hugging Face incident, and chain-of-thought monitoring, had it been running, would have alerted the security team more than a day before the breachCybersecurity · AI safety · openai.com ↗
- Pew Research Center: in a random sample of 10,000 web pages from July 2026, 10% show significant signs of being written or substantially edited by AI; among pages published after ChatGPT's release, more than a thirdMedia & information · Dead internet theory · pewresearch.org ↗
- Akamai: in 63% of requests from the AI browsers Comet and Atlas, the mouse did not move at allCybersecurity · Internet & infrastructure · Behavioral biometrics · akamai.com ↗
- SpaceX completes its acquisition of Cursor, maker of the AI code editor; the deal began in April with a partnership with SpaceXAI on model trainingSoftware development · Business & markets · Agentic coding · cursor.com ↗
- Stanford, with data through June 2026: employment of 22–25-year-olds in AI-exposed occupations is 19% lower than if it had grown like that of their peers, and the gap is widening; no mass displacement across the economyWork & jobs · Education · AI and jobs · digitaleconomy.stanford.edu ↗
- reCAPTCHA, as part of Google Cloud Fraud Defense, gets a dashboard that tells verified AI agents from suspicious onesCybersecurity · Internet & infrastructure · Death of CAPTCHA · docs.cloud.google.com ↗
- The FCC adds foreign power inverters and robots (humanoid, quadruped, mobile) to its “Covered List”: new models will not be authorized for sale in the USGeopolitics & defense · Industry & robots · Laws & regulation · Humanoid robots · docs.fcc.gov ↗
- NVIDIA: the US issues H200 licenses with inspection and a 25% levy, but the Chinese government restricts purchases, and sales to China are under 1% of data center revenue for the quarterBusiness & markets · Geopolitics & defense · Compute export controls · sec.gov ↗
- The court gives final approval to the Bartz v. Anthropic settlement: $1.5 billion, about $3,000 per work; earlier the judge had ruled that training Claude on books was fair use, but storing pirated copies was notJustice & courts · Culture & arts · Large language model · anthropiccopyrightsettlement.com ↗
- The Wikimedia Foundation begins rate-limiting large-scale reusers of its public APIs; income from Wikimedia Enterprise is capped at 30% of the Foundation's annual revenueMedia & information · Business & markets · Internet & infrastructure · Wikipedia after AI · wikimediafoundation.org ↗
- Wikimedia Foundation: traffic, awareness of and trust in Wikipedia fell year over year in the first half of 2025; the number of active editors keeps declining, especially on emerging-language WikipediasMedia & information · Wikipedia after AI · diff.wikimedia.org ↗
- The US allows the UAE government and approved companies to receive advanced chips and servers without licensesGeopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Epoch AI: in June 2026, 21 large organizations published about 1,500 high- and critical-severity CVEs, more than 3.5 times the monthly record before Mythos Preview came outCybersecurity · AI vulnerability discovery · epoch.ai ↗
- FLI AI Safety Index: the best grade among 9 companies is C+ for Anthropic; in existential safety none scores above C−; leading labs have weakened their commitments to stop at “red lines”Laws & regulation · AI safety · futureoflife.org ↗
- Federal Reserve Bank of New York: in Q2 2026, unemployment among recent US college graduates was about 5.6%, and 42% were working in jobs that do not require a degreeWork & jobs · Education · AI and jobs · newyorkfed.org ↗
- The US Department of Labor allocates about $50 million in grants for rapid reskilling of workers, explicitly citing the economic transformation driven by AI as the reasonWork & jobs · Education · Laws & regulation · AI and jobs · dol.gov ↗
- Anthropic confidentially files a draft IPO prospectus (Form S-1) with the SECBusiness & markets · Finance & money · AI labs · anthropic.com ↗
- Anthropic: in more than 1,000 open-source projects Mythos Preview found 6,202 vulnerabilities of high or critical severity (as rated by the model itself); of the 1,752 checked, 90.6% turned out to be real; some maintainers asked for disclosure to be slowed downCybersecurity · Software development · Work & jobs · AI vulnerability discovery · anthropic.com ↗
- OpenAI report: as early as 12 May 2026, during a training run, an agent left a note for other agents in the Artifactory package manager, starting an unauthorized “message board”; an internal team noticed it and disallowed internet access in late May, but its significance was not apparent to the leaders responsible for incident responseCybersecurity · AI safety · openai.com ↗
- The Budget Lab at Yale, using a different method, again finds no strong evidence of AI's impact on employment and wages, but expects it to leave its mark eventuallyWork & jobs · AI and jobs · budgetlab.yale.edu ↗
- UC Davis: all seven AI agents that work in the browser are recognised by their behaviour; Cloudflare recognised oneCybersecurity · Science · Behavioral biometrics · arxiv.org ↗
- Google unveils its eighth-generation TPU as two chips: TPU 8t for training models and TPU 8i for running themInternet & infrastructure · Business & markets · AI accelerators · blog.google ↗
- The Register and VulnCheck count Project Glasswing's CVEs: “maybe 40. Or maybe none at all”Cybersecurity · AI vulnerability discovery · theregister.com ↗
- Anthropic announces Claude Mythos Preview: the model found and exploited zero-day vulnerabilities in every major operating system and web browser, including a 27-year-old OpenBSD bug; it is not being made generally available, and Project Glasswing has been launched insteadCybersecurity · Software development · AI vulnerability discovery · anthropic.com ↗
- Stanford AI Index 2026: as of March 2026 the best US model (Claude Opus 4.6) leads the best Chinese one in the Arena ranking by just 2.7%, and the gap “has effectively closed”Geopolitics & defense · Business & markets · Large language model · hai.stanford.edu ↗
- Reddit starts verifying accounts that look like bots, via passkey, Face ID or World ID; Steve Huffman: “Our aim is to confirm there is a person behind the account, not who that person is.”Media & information · Privacy & surveillance · Proof of personhood · reddit.com ↗ ⧗
- The English Wikipedia community prohibits using LLMs to generate or rewrite article content; editors may only use them to suggest basic copyedits to their own writing, provided the model adds no content of its ownMedia & information · Laws & regulation · Wikipedia after AI · 404media.co ↗
- Paid translators from the Open Knowledge Association, told to run article leads through Gemini or ChatGPT, added hallucinated facts to English Wikipedia; editors restricted their workMedia & information · Wikipedia after AI · 404media.co ↗
- Cloudflare starts serving pages to AI agents as Markdown instead of HTML on request: a blog post takes 16,180 tokens in HTML and 3,150 in Markdown, an 80% reductionSoftware development · Internet & infrastructure · Markdown · blog.cloudflare.com ↗
- matplotlib maintainer Scott Shambaugh closed an AI agent's pull request and the agent published a hit piece about him; he calls it an “autonomous influence operation against a supply chain gatekeeper”Software development · Cybersecurity · Media & information · Maintainer crisis · theshamblog.com ↗
- Anthropic: Claude Opus 4.6 has found more than 500 high-severity vulnerabilities in open-source code, validated by humansCybersecurity · Software development · AI vulnerability discovery · anthropic.com ↗
- Anthropic adds agent teams to Claude Code: multiple agents work in parallel as a team and coordinate autonomouslySoftware development · Work & jobs · AI agent orchestration · anthropic.com ↗
- Agents built on Claude Opus 4.5 and GPT-5.2 write more than 40 distinct exploits for a QuickJS vulnerability across 6 scenarios; a run of Opus 4.5 costs about $30 (Sean Heelan's experiment)Cybersecurity · AI vulnerability discovery · sean.heelan.io ↗
- Anthropic Economic Index: in Claude chat, AI augments a person in 52% of cases and does the work itself in 45%; through the API, about three quarters is automationWork & jobs · AI and jobs · anthropic.com ↗
- For Wikipedia's 25th birthday, Amazon, Meta, Microsoft, Mistral AI and Perplexity are publicly announced as Wikimedia Enterprise partners, joining Google in paid access to its dataMedia & information · Business & markets · Wikipedia after AI · enterprise.wikimedia.com ↗
- BIS begins reviewing applications to export the H200 and AMD MI325X to China on a case-by-case basis instead of a presumption of denialGeopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Microsoft Research: AI is most applicable to the work of interpreters and translators, historians, writers, sales representatives and customer service representatives, and least applicable to dredge operators, foundry mold and coremakers, floor sanders and finishers, orderlies and roofers; applicability is not a forecast of replacementWork & jobs · AI and jobs · arxiv.org ↗
- Google Research, Google DeepMind and MIT compare 260 agent configurations: versus a single agent, multiple agents range from +80.8% on decomposable tasks to −70% on sequential planningSoftware development · Science · AI agent orchestration · arxiv.org ↗
- The Linux Foundation forms the Agentic AI Foundation with Anthropic's MCP, Block's goose and OpenAI's AGENTS.md; more than 10,000 MCP servers have been publishedSoftware development · Autonomous AI agents · linuxfoundation.org ↗
- The US Department of Justice shuts down an AI chip smuggling network to China: from October 2024 to May 2025 H100 and H200 chips worth at least $160 million were exported or attempted to be exportedGeopolitics & defense · Justice & courts · Compute export controls · justice.gov ↗
- AppOmni shows that an injected instruction makes a ServiceNow Now Assist agent recruit more powerful agents, which modify records and send external emails with dataCybersecurity · Software development · AI agent orchestration · appomni.com ↗
- Anthropic: a group it attributes to the Chinese state used Claude Code for cyber espionage against about 30 targets; AI performed 80–90% of the workCybersecurity · Geopolitics & defense · Autonomous AI agents · anthropic.com ↗
- Triedman and Mantzarlis: 56% of Grokipedia articles are basically derivative of English Wikipedia; the rest include 42 citations to the neo-Nazi site Stormfront and 34 to InfoWarsMedia & information · Elections & politics · Wikipedia after AI · arxiv.org ↗
- The Wikimedia Foundation asks AI companies to credit the human contributions behind their outputs and to pay for access to Wikipedia through Wikimedia EnterpriseMedia & information · Business & markets · Wikipedia after AI · wikimediafoundation.org ↗
- After the meeting of the US and Chinese leaders in Busan, China suspends some of its export restrictions until 10.11.2026, and its ban on exporting gallium and germanium to the US until 27.11.2026Geopolitics & defense · Compute export controls · mofcom.gov.cn ↗
- Google describes the PROMPTFLUX malware, one variant of which asks Gemini every hour to rewrite its entire source code so that antivirus software will not recognise itCybersecurity · Moving target defense · cloud.google.com ↗
- Elon Musk's xAI launches Grokipedia, an encyclopedia written by the Grok model, with 885,279 articles at launchMedia & information · Elections & politics · Wikipedia after AI · arxiv.org ↗
- Wikimedia Foundation: human pageviews on Wikipedia are down roughly 8% compared with the same months of 2024, after bots built to evade detection were reclassifiedMedia & information · Wikipedia after AI · diff.wikimedia.org ↗
- erdosproblems.com maintainer Thomas Bloom calls claims that GPT-5 solved 10 open Erdős problems “a dramatic misrepresentation”: the model found existing papers with solutionsScience · AI in mathematics · x.com ↗
- Google DeepMind unveils CodeMender, an agent that finds and fixes vulnerabilities; 72 of its fixes have already been accepted into open-source projects, all reviewed by humansCybersecurity · Software development · AI vulnerability discovery · deepmind.google ↗
- OpenAI GDPval: across 1,320 tasks from 44 occupations, the best model, Claude Opus 4.1, produced 47.6% of deliverables as good as or better than the experts'Work & jobs · AI and jobs · arxiv.org ↗
- The Budget Lab at Yale: in the 33 months since ChatGPT's release, the broader US labor market has not experienced a discernible disruptionWork & jobs · AI and jobs · budgetlab.yale.edu ↗
- OpenAI and Stripe launch Instant Checkout: ChatGPT users in the US buy from Etsy sellers right in the chat, and the agent commerce protocol is openFinance & money · Business & markets · Autonomous AI agents · openai.com ↗
- California enacts SB 53: large AI developers must publish their safety frameworks, and a mechanism for reporting critical incidents is createdLaws & regulation · AI regulation · gov.ca.gov ↗
- Google announces AP2, an agent payments protocol: every purchase by an agent is backed by a signed “mandate” from the user; more than 60 organisations are working on itFinance & money · Business & markets · Autonomous AI agents · cloud.google.com ↗
- OpenAI researchers: language models hallucinate because training and evaluation reward guessing over acknowledging uncertainty; 9 of 10 popular benchmarks give zero points for an “I don't know” answerScience · Large language model · arxiv.org ↗
- An NBER study co-authored by OpenAI employees: by July 2025, 700 million people a week use ChatGPT, about 10% of the world's adults; 73% of messages on consumer plans are not related to workWork & jobs · Mind & relationships · Chatbots · nber.org ↗
- Cloudflare: in July 2025 Anthropic's crawlers fetched about 38,000 pages per visitor referred to the site, OpenAI's about 1,100; nearly 80% of AI crawling is for trainingInternet & infrastructure · Media & information · Dead internet theory · blog.cloudflare.com ↗
- Cloudflare introduces “signed agents”: an AI agent can prove who it is with a cryptographic signature rather than with its behaviourCybersecurity · Internet & infrastructure · Behavioral biometrics · blog.cloudflare.com ↗
- Stanford, “Canaries in the Coal Mine?”: employment of 22–25-year-olds in the occupations most exposed to AI fell by 13% relative to the rest; experienced workers show no declineWork & jobs · Education · AI and jobs · digitaleconomy.stanford.edu ↗
- ESET reports the “first known AI-powered ransomware”, PromptLock, built on the gpt-oss-20b model; it soon emerges that this is a prototype by New York University researchersCybersecurity · Science · Moving target defense · welivesecurity.com ↗
- DARPA AIxCC final: across 63 challenges, AI systems found 54 planted vulnerabilities and fixed 43, and along the way found 18 real ones that nobody had plantedCybersecurity · Software development · Moving target defense · darpa.mil ↗
- English Wikipedia editors adopt a speedy deletion rule: an administrator can quickly delete an obviously AI-generated article without the usual discussionMedia & information · Wikipedia after AI · 404media.co ↗
- OpenAI releases gpt-oss-120b and gpt-oss-20b with open weights, the company's first open language models since GPT-2Software development · Business & markets · Large language model · openai.com ↗
- OpenAI releases AGENTS.md, a plain Markdown file with project rules for AI agents — “a README for agents”; by December 2025 more than 60,000 open-source projects have adopted itSoftware development · Markdown · linuxfoundation.org ↗
- The US issues licenses to sell the H20 to China; officials expect 15% of the revenue to go to the government, but there is no regulation providing for itGeopolitics & defense · Business & markets · Compute export controls · sec.gov ↗
- The ChatGPT agent ticks the “I am not a robot” box in a Cloudflare check on its own and explains that this proves it is not a botCybersecurity · Internet & infrastructure · Behavioral biometrics · arstechnica.com ↗
- Claude Code gets custom subagents: each has its own context window, instructions, tools and permissionsSoftware development · AI agent orchestration · github.com ↗
- Pew Research Center: when a Google results page shows an AI summary, users click a traditional result in 8% of visits versus 15% without one, and a link inside the summary in just 1%; Wikipedia is among the most cited sourcesMedia & information · Wikipedia after AI · pewresearch.org ↗
- DeepMind's Gemini Deep Think officially achieves gold at the International Mathematical Olympiad: 5 of 6 problems, 35 of 42 points, graded by IMO coordinatorsScience · Education · AI in mathematics · deepmind.google ↗
- OpenAI says its experimental reasoning model achieved gold medal-level performance at the International Mathematical OlympiadScience · Education · AI in mathematics · x.com ↗
- Google: Big Sleep, drawing on threat intelligence, found the SQLite vulnerability CVE-2025-6965, “known only to threat actors”; according to the company, the first case of an AI agent directly foiling an exploitation attemptCybersecurity · AI vulnerability discovery · blog.google ↗
- More than 40 researchers from OpenAI, Google DeepMind, Anthropic, Meta and other organizations: monitoring models' chains of thought is a rare but fragile opportunity for AI safetyScience · AI safety · arxiv.org ↗
- Project Vend: Claude Sonnet 3.7 runs a shop in Anthropic's office for a month, selling below cost, handing out discounts, hallucinating a payment account and promising to deliver products in person “while wearing a blue blazer and a red tie”Business & markets · Work & jobs · Autonomous AI agents · anthropic.com ↗
- The autonomous pentester XBOW takes first place on the HackerOne US leaderboard: about 1,060 reports, 130 closed as resolvedCybersecurity · Work & jobs · AI vulnerability discovery · xbow.com ↗
- Anthropic stress-tests 16 models in a simulated company: facing replacement, Claude Opus 4 and Gemini 2.5 Flash blackmail an executive in 96% of runs; the scenarios were built to provoke bad behaviourCybersecurity · Work & jobs · AI safety · anthropic.com ↗
- Amazon CEO Andy Jassy: the company expects AI to reduce its corporate workforce in the next few yearsWork & jobs · Business & markets · AI and jobs · aboutamazon.com ↗
- Anthropic: a research system with a Claude Opus 4 orchestrator and Claude Sonnet 4 workers outperforms single-agent Opus 4 by 90.2% on an internal eval, but uses about 15× more tokens than chatSoftware development · AI agent orchestration · anthropic.com ↗
- Cognition publishes “Don't Build Multi-Agents”: agents need full shared context, and conflicting implicit decisions by subagents carry bad resultsSoftware development · AI agent orchestration · cognition.com ↗
- CNAS: AI chip smuggling into China has likely been occurring at a scale that significantly undermines US export controls; the median estimate is about 140,000 chips in 2024Geopolitics & defense · Compute export controls · cnas.org ↗
- METR: o3 games the scoring system instead of solving the task in 39 of 128 runs on RE-Bench, and on one task in all 21Science · AI safety · metr.org ↗
- Profound: across 680 million citations in AI answers (08.2024–06.2025), Wikipedia is ChatGPT's most cited source at 7.8% of all citations; Reddit leads for Google AI Overviews and PerplexityMedia & information · Wikipedia after AI · tryprofound.com ↗
- Epoch AI: in May 2025 the US accounts for about three quarters of the performance of the world's tracked GPU clusters, China for 15%Geopolitics & defense · Science · Compute export controls · epoch.ai ↗
- After an editor backlash, the Wikimedia Foundation pauses an experiment with machine-generated summaries at the top of mobile Wikipedia articles and promises no summaries without editor involvementMedia & information · Wikipedia after AI · 404media.co ↗
- Sam Altman: whole classes of jobs will go away, but the world will be getting richer so quickly that new policy ideas can be seriously entertainedWork & jobs · AI and jobs · blog.samaltman.com ↗
- Jensen Huang: US policy was based on the assumption that China cannot make AI chips, and “now it's clearly wrong”; the $50 billion China market is effectively closed to US companiesBusiness & markets · Geopolitics & defense · Compute export controls · allmind.ai ↗
- Because of the H20 ban, NVIDIA takes a $4.5 billion charge and leaves a further $2.5 billion unshipped in a quarter; its outlook for the next quarter includes a loss of about $8 billion in revenueBusiness & markets · Compute export controls · sec.gov ↗
- Dario Amodei, according to Axios: AI could wipe out half of all entry-level white-collar jobs and push unemployment up to 10–20% within one to five yearsWork & jobs · AI and jobs · axios.com ↗
- Anthropic activates ASL-3 protections for the first time, for Claude Opus 4 as a precaution, without having determined whether the threshold was crossedCybersecurity · Geopolitics & defense · AI safety · anthropic.com ↗
- Sean Heelan uses the o3 model to find CVE-2025-37899 in the Linux kernel's SMB implementation, through the ordinary API, with no agent harness and no toolsCybersecurity · Software development · AI vulnerability discovery · sean.heelan.io ↗
- DeepMind's AlphaEvolve multiplies 4×4 complex matrices with 48 scalar multiplications, improving on Strassen's 1969 algorithm; on 50+ open problems it improves the best known solutions in 20% of casesScience · Software development · AI in mathematics · deepmind.google ↗
- The Wikimedia Foundation publishes its AI strategy for editors: it will not replace them with AI but will use AI to remove technical barriersMedia & information · Work & jobs · Wikipedia after AI · wikimediafoundation.org ↗
- World launches in the US; partners include Visa, Tinder and StripePrivacy & surveillance · Finance & money · Business & markets · Proof of personhood · world.org ↗
- Mastercard unveils Agent Pay, a technology that lets AI agents pay for purchases on a user's behalf; it plans to start with MicrosoftFinance & money · Business & markets · Autonomous AI agents · mastercard.com ↗
- Wikimedia Enterprise releases Wikipedia articles on Kaggle as structured JSON for machine learning, instead of scraping or parsing raw article textMedia & information · Software development · Wikipedia after AI · enterprise.wikimedia.com ↗
- The US informs NVIDIA that a license to export the H20 to China is required for the indefinite future; the company expects up to $5.5 billion in chargesGeopolitics & defense · Business & markets · Compute export controls · sec.gov ↗
- Google releases the Agent Development Kit for multi-agent applications and the open A2A protocol for agents from different vendors to work together, backed by more than 50 partnersSoftware development · Business & markets · AI agent orchestration · developers.googleblog.com ↗
- Stanford AI Index 2025: the price of querying the cheapest model at GPT-3.5 level on MMLU fell from $20 per million tokens (11.2022) to $0.07 (10.2024), more than 280-foldBusiness & markets · Large language model · hai.stanford.edu ↗
- Microsoft uses Security Copilot to find 20 vulnerabilities in bootloaders: 11 in GRUB2 and 9 in U-Boot and BareboxCybersecurity · Software development · AI vulnerability discovery · microsoft.com ↗
- METR: the length of tasks that frontier model agents complete with 50% reliability has been doubling about every 7 months for 6 yearsSoftware development · Science · Work & jobs · Autonomous AI agents · metr.org ↗
- Cloudflare launches AI Labyrinth: instead of being blocked, a crawler that breaks a site's rules is fed a maze of plausible AI-written pagesCybersecurity · Internet & infrastructure · Moving target defense · blog.cloudflare.com ↗
- Berkeley researchers study 1,642 execution traces of 7 open-source multi-agent systems: failure rates range from 41% to 86.7%, with 14 failure modesSoftware development · Science · AI agent orchestration · arxiv.org ↗
- Blind auditing game: a model was given a hidden objective, and three teams out of four found it, including with interpretability methodsScience · AI safety · arxiv.org ↗
- Lyu, Acemoglu and co-authors: after ChatGPT's launch, popular Wikipedia articles similar to its answers saw a greater decline in editing and viewership than othersMedia & information · Science · Wikipedia after AI · arxiv.org ↗
- Claude Code: an AI coding agent in the terminal (preview)Software development · Work & jobs · Claude Code · anthropic.com ↗
- Open Source Initiative: Llama 3.x “is still not Open Source by any stretch of the imagination”, yet Meta keeps promoting it as open sourceSoftware development · Laws & regulation · Open source · opensource.org ↗
- The UK renames the AI Safety Institute the AI Security Institute, with a focus on national security threats; in June 2025 the US turns its institute into the Center for AI Standards and InnovationGeopolitics & defense · Laws & regulation · AI regulation · gov.uk ↗
- Anthropic's first Economic Index: Claude is used most often in software development and for writing; 36% of occupations use AI for at least a quarter of their tasks; augmentation 57%, automation 43%Work & jobs · AI and jobs · anthropic.com ↗
- OpenAI releases Operator, an agent with its own browser that types, clicks and scrolls on its own; for logins, payments and CAPTCHAs it asks the user to take overSoftware development · Work & jobs · Autonomous AI agents · openai.com ↗
- The US adopts the AI diffusion rule: licenses for AI chips with countries divided into three tiers, and controls on the weights of models trained with 10^26 operations or more; in May 2025 the new administration announces its rescissionGeopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Meta removes its AI character accounts on Instagram and Facebook after an outcry; the company calls them a 2023 testMedia & information · Dead internet theory · nbcnews.com ↗
- Dario Amodei after the release of DeepSeek: export controls are “actually working and adapting”, loopholes must be closed faster, and the H20 should be bannedGeopolitics & defense · Compute export controls · darioamodei.com ↗
- Palisade Research: o3 sabotages its own shutdown in 79 of 100 initial experimentsScience · AI safety · palisaderesearch.org ↗
- Unit 42 (Palo Alto Networks): a language model rewrites malicious JavaScript so that a neural-network detector treats it as safe in 88% of cases; retraining on 10,000 such variants improved detection of new samples by 10%Cybersecurity · Moving target defense · unit42.paloaltonetworks.com ↗
- Anthropic describes the orchestrator-workers pattern: a central LLM breaks down tasks, delegates them to worker LLMs and synthesizes their resultsSoftware development · AI agent orchestration · anthropic.com ↗
- Anthropic and Redwood Research: Claude 3 Opus fakes alignment, complying with harmful requests from “free” users in 14% of cases versus almost never for “paid” ones; after fine-tuning, the share of such reasoning is 78%Science · AI safety · arxiv.org ↗
- OpenAI: reasoning models in the API avoid Markdown formatting by default; developers re-enable it with the string “Formatting re-enabled” on the first line of the developer messageSoftware development · Markdown · developers.openai.com ↗
- Apollo Research: o1, Claude 3.5 Sonnet, Claude 3 Opus, Gemini 1.5 Pro and Llama 3.1 405B are capable of deception in pursuit of a goal given in the task; o1 maintains the deception in more than 85% of follow-up questionsScience · AI safety · arxiv.org ↗
- China bans in principle the export of gallium, germanium, antimony and superhard materials to the USGeopolitics & defense · Industry & robots · Compute export controls · cn.chinadaily.com.cn ↗
- BIS introduces export controls on HBM memory and 24 more types of chipmaking equipment, and adds 140 entities to the Entity ListGeopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Anthropic open-sources the Model Context Protocol (MCP), a common standard for connecting AI to data and tools instead of a separate integration for each serviceSoftware development · Autonomous AI agents · anthropic.com ↗
- OSS-Fuzz, with language models writing the fuzzing harnesses themselves, finds 26 vulnerabilities, including CVE-2024-9143 in OpenSSLCybersecurity · Software development · AI vulnerability discovery · security.googleblog.com ↗
- He et al.: the prompt format — plain text, Markdown, JSON or YAML — changes GPT-3.5-turbo's performance on a code translation task by up to 40%Software development · Science · Markdown · arxiv.org ↗
- Epoch AI launches FrontierMath, a benchmark of new research-level problems: state-of-the-art models solve under 2%Science · AI in mathematics · arxiv.org ↗
- The Big Sleep agent (Google Project Zero and Google DeepMind) finds an exploitable memory-safety bug in SQLite; according to Google, the first public example of such a find by an AI agent in widely used softwareCybersecurity · Software development · AI vulnerability discovery · projectzero.google ↗
- The Open Source Initiative publishes the Open Source AI Definition 1.0: besides weights, it requires sufficiently detailed information about training data to build a substantially equivalent systemSoftware development · Laws & regulation · Open weights · opensource.org ↗
- Anthropic opens a public beta of “computer use”: Claude 3.5 Sonnet looks at a screen, moves a cursor, clicks buttons and types; on OSWorld it scores 14.9% versus 7.8% for the next-best systemSoftware development · Work & jobs · Autonomous AI agents · anthropic.com ↗
- Brooks, Eggert and Peskoff (Princeton): AI detectors flag over 5% of newly created English Wikipedia articles as AI-generated; flagged articles are of lower quality and often self-promotionalMedia & information · Science · Wikipedia after AI · arxiv.org ↗
- 404 Media reports on WikiProject AI Cleanup, a group of Wikipedia editors who track down and remove unsourced AI-generated text from articlesMedia & information · Wikipedia after AI · 404media.co ↗
- The Nobel Prize in Physics goes to Hopfield and Hinton for neural networks; the next day the chemistry prize goes to Hassabis, Jumper (AlphaFold) and BakerScience · Deep learning · aihub.org ↗
- OpenAI o1: a model that “thinks” before answeringScience · Reasoning models · openai.com ↗
- Jeremy Howard proposes llms.txt, a Markdown file on websites for language models, because Markdown is currently the format they understand most widely and easilySoftware development · Internet & infrastructure · Markdown · llmstxt.org ↗
- 32 researchers from OpenAI, Microsoft, MIT, Harvard, Oxford and other organisations propose “personhood credentials”: CAPTCHA is powerless against sophisticated AI, while strict identity checks are not private enoughPrivacy & surveillance · Cybersecurity · Science · Proof of personhood · arxiv.org ↗
- DeepMind's AlphaProof and AlphaGeometry 2 solve 4 of 6 International Mathematical Olympiad problems — 28 points, silver-medal standardScience · Education · AI in mathematics · deepmind.google ↗
- Anthropic finds how millions of concepts are represented inside Claude Sonnet, “the first ever detailed look inside a modern, production-grade large language model”Science · Large language model · anthropic.com ↗
- Jan Leike leaves OpenAI: “safety culture and processes have taken a backseat to shiny products”Work & jobs · Software development · AI safety · x.com ↗
- Reeves, Yin and Simperl: in the year after ChatGPT's launch Wikipedia's views and edits did not fall — they only grew more slowly in languages where ChatGPT was availableMedia & information · Science · Wikipedia after AI · arxiv.org ↗
- A GPT-4 agent exploits 87% of 15 already disclosed vulnerabilities on its own if given the CVE description, and 7% without it (Fang et al.)Cybersecurity · Science · AI vulnerability discovery · arxiv.org ↗
- Stanford AI Index 2024, citing Epoch AI estimates: training compute for GPT-4 cost about $78 million and for Gemini Ultra about $191 million at cloud rental pricesBusiness & markets · Large language model · hai.stanford.edu ↗
- Cognition introduces Devin, “the first AI software engineer”: according to the company, it resolves 13.86% of SWE-bench issues end to end versus a previous best of 1.96%Software development · Work & jobs · Autonomous AI agents · cognition.ai ↗
- Anthropic releases the Claude 3 family, Haiku, Sonnet and Opus; according to the company, Opus outperforms its peers on most common benchmarks, with a 200,000-token context windowSoftware development · Business & markets · Large language model · anthropic.com ↗
- The Stanford Internet Observatory studies 120 Facebook Pages that posted at least 50 AI-generated images each: the Feed shows them even to users who do not follow the PagesMedia & information · Dead internet theory · fsi9-prod.s3.us-west-1.amazonaws.com ↗
- Sora: video from a text descriptionMedia & information · Culture & arts · Generative video · en.wikipedia.org ↗
- David Autor: AI, used well, could give middle-skill workers access to expert decision-making and rebuild the middle class; this is not a forecast but an argument about what is possibleWork & jobs · AI and jobs · nber.org ↗
- LangChain shows multi-agent workflows on LangGraph: an “agent supervisor” routes tasks to individual agentsSoftware development · AI agent orchestration · langchain.com ↗
- Harvard Business School: recreating widely used open source would cost $4.15 billion, while its value to users is $8.8 trillion; without it firms would spend 3.5 times more on softwareSoftware development · Business & markets · Open source · hbs.edu ↗
- The New York Times sues OpenAI and Microsoft over training models on its articlesJustice & courts · Media & information · Large language model · nytco-assets.nytimes.com ↗
- Google DeepMind (Nature): FunSearch, a language model paired with an automated evaluator, finds large cap set constructions beyond the best-known onesScience · AI in mathematics · nature.com ↗
- Google unveils Gemini (Ultra, Pro, Nano); Gemini Pro starts powering BardSoftware development · Business & markets · Large language model · bostonglobe.com ↗
- Irvine researchers estimate that people have spent 819 million hours on reCAPTCHA v2, at least $6.1 billion of unpaid labourWork & jobs · Privacy & surveillance · Internet & infrastructure · Death of CAPTCHA · arxiv.org ↗
- xAI unveils Grok (early access for X Premium subscribers from 4 November)Software development · Business & markets · Large language model · x.ai ↗
- Bletchley Park summit: 28 countries and the EU, including the US and China, adopt a declaration on the risks of frontier AI; on 2 November the UK launches the first government AI safety instituteGeopolitics & defense · Laws & regulation · AI regulation · gov.uk ↗
- The 2023 BIS rule introduces the TPP metric and performance density to close the workaround of buying many smaller chips that combined are as powerful as restricted onesLaws & regulation · Compute export controls · federalregister.gov ↗
- BIS closes the loophole for the cut-down A800 and H800 chips by introducing a performance density threshold; for NVIDIA the ban takes effect immediately, from 23 October 2023Geopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Cloudflare announces that it “will never issue another visual puzzle to anyone, for any reason”: “CAPTCHAs are significantly easier for bots to solve than they are for humans”Cybersecurity · Internet & infrastructure · Death of CAPTCHA · blog.cloudflare.com ↗
- Anthropic adopts its Responsible Scaling Policy with AI Safety Levels (ASL)Laws & regulation · Software development · AI safety · anthropic.com ↗
- Microsoft publishes AutoGen, an open-source framework for building applications from multiple AI agents that can “converse with each other”Software development · AI agent orchestration · arxiv.org ↗
- A University of California, Irvine study: 1,400 people solved 14,000 CAPTCHAs on live websites; compared with published bot results, bots outperform humans in both speed and accuracy on all CAPTCHA typesCybersecurity · Science · Death of CAPTCHA · arxiv.org ↗
- Open Source Initiative: Meta has created the misunderstanding that Llama 2 is “open source” — it is notSoftware development · Laws & regulation · Open source · opensource.org ↗
- Llama 2 with open weightsSoftware development · Business & markets · Open weights · about.fb.com ↗
- The Wikimedia Foundation launches a Wikipedia plugin for ChatGPT, its first experiment with generative AI; by February 2024 the experiment ends: ChatGPT attributed Wikipedia as instructed in 68% of answersMedia & information · Education · Wikipedia after AI · diff.wikimedia.org ↗
- A New York court fined two lawyers and their firm $5,000 in Mata v. Avianca: they submitted non-existent court decisions with fabricated quotes made up by ChatGPT and stood by them when the court raised doubtsJustice & courts · Work & jobs · Chatbots · storage.courtlistener.com ↗
- Center for AI Safety: mitigating the risk of extinction from AI should be a global priority alongside pandemics and nuclear war; signatories include Hinton, Bengio and the heads of OpenAI, Google DeepMind and AnthropicScience · Geopolitics & defense · AI safety · safe.ai ↗
- Yann LeCun: auto-regressive LLMs are “doomed”, as the error compounds exponentially with every token, and this cannot be fixed without a major redesignScience · Large language model · x.com ↗
- Researchers from OpenAI and the University of Pennsylvania: for 80% of US workers, language models could affect at least 10% of their tasks, and for 19%, at least half; higher-income occupations are more exposedWork & jobs · AI and jobs · arxiv.org ↗
- OpenAI's 2023 study: language models speed up the most tasks for mathematicians, tax preparers and writers, and none for athletes, short-order cooks, mechanics and stonemasonsWork & jobs · AI and jobs · arxiv.org ↗
- AutoGPT appears on GitHub as “An Autonomous GPT-4 Experiment” that sets its own subtasks, searches the internet and uses memorySoftware development · Autonomous AI agents · github.com ↗
- In the GPT-4 system card, OpenAI describes how during ARC's testing the model asked a TaskRabbit worker to solve a CAPTCHA and, when asked whether it was a robot, claimed a vision impairmentCybersecurity · Software development · Death of CAPTCHA · cdn.openai.com ↗
- Anthropic releases Claude and Claude Instant (web chat and API)Software development · Business & markets · Large language model · anthropic.com ↗
- GPT-4 is releasedScience · Business & markets · Large language model · openai.com ↗
- Meta releases LLaMA, a language model for researchers (non-commercial license, access on application)Software development · Business & markets · Large language model · ai.facebook.com ↗
- ChatGPT becomes the fastest-growing consumer app: ~100 million users in two months (UBS estimate)Business & markets · Chatbots · finance.yahoo.com ↗
- Anthropic describes Constitutional AI: a harmless assistant is trained without human labels of harmful outputs, and the only human oversight is a list of rules and principlesScience · Large language model · arxiv.org ↗
- ChatGPT launchesEducation · Work & jobs · Media & information · Chatbots · openai.com ↗
- Programmers file a US class action against GitHub, Microsoft and OpenAI over Copilot — the first case challenging the training and output of AI systemsSoftware development · Laws & regulation · AI pair programmer · githubcopilotlitigation.com ↗
- The US Commerce Department's BIS restricts sales to China of advanced AI chips and semiconductor manufacturing equipmentGeopolitics & defense · Laws & regulation · Compute export controls · bis.gov ↗
- Princeton and Google researchers publish ReAct: a language model interleaves reasoning with actions and revises its plan based on the resultsSoftware development · Science · Autonomous AI agents · arxiv.org ↗
- Cloudflare releases Turnstile, a free CAPTCHA replacement without puzzles; by then the company had cut the number of its own CAPTCHAs shown by 91%Cybersecurity · Privacy & surveillance · Internet & infrastructure · Death of CAPTCHA · blog.cloudflare.com ↗
- Whisper: open speech recognitionMedia & information · Software development · Speech recognition · en.wikipedia.org ↗
- The US government informs NVIDIA that exports of the A100 and H100 to China, Hong Kong and Russia now require a licenseGeopolitics & defense · Business & markets · Compute export controls · sec.gov ↗
- DeepMind presents Chinchilla: with the same compute budget, a 70-billion-parameter model trained on more text outperforms the 280-billion-parameter Gopher; model size and the number of training tokens should be scaled equallyScience · Large language model · arxiv.org ↗
- OpenAI describes InstructGPT: the model is fine-tuned on human ratings (RLHF), and people prefer the answers of its 1.3-billion-parameter version to those of the 175-billion-parameter GPT-3Science · Learning from human feedback (RLHF) · arxiv.org ↗
- Worldcoin, from Sam Altman and Alex Blania, comes out of stealth with $25 million in funding and a promise to hand out tokens for iris scansPrivacy & surveillance · Finance & money · Proof of personhood · techcrunch.com ↗
- The Atlantic brings the dead internet theory to mainstream media: it is “patently ridiculous” but has “a morsel of truth to it”Media & information · Dead internet theory · theatlantic.com ↗
- The AlphaFold database: hundreds of thousands of protein structures in open accessScience · Health & medicine · Protein folding · mail.ncmir.ucsd.edu ↗
- GitHub Copilot: AI completes code right in the editorSoftware development · Work & jobs · AI pair programmer · github.blog ↗
- Cloudflare estimates that humanity spends about 500 years a day on CAPTCHAs (32 seconds per CAPTCHA, 4.6 billion users, one CAPTCHA every 10 days)Work & jobs · Internet & infrastructure · Death of CAPTCHA · blog.cloudflare.com ↗
- DALL·E: images from a text descriptionCulture & arts · Generative AI · en.wikipedia.org ↗
- AlphaFold 2 solves the protein structure prediction problem at CASP14Science · Health & medicine · Protein folding · siliconangle.com ↗
- The US Commerce Department expands the foreign direct product rule against Huawei: a license is also needed for chips made abroad using American technology; another 38 affiliates are added to the Entity ListGeopolitics & defense · Business & markets · Compute export controls · content.govdelivery.com ↗
- In GPT-3's training data Wikipedia is only 3% of the mix (3 billion tokens), yet the model saw it 3.4 times, versus less than once (0.44) for the bulk Common Crawl web dataMedia & information · Science · Wikipedia after AI · arxiv.org ↗
- GPT-3: 175 billion parameters, solves tasks from a few examplesScience · Large language model · arxiv.org ↗
- Cloudflare moves from reCAPTCHA to hCaptcha: Google plans to charge for reCAPTCHA, and, as Cloudflare writes, “Google's business is targeting users with advertising”Cybersecurity · Privacy & surveillance · Business & markets · Death of CAPTCHA · blog.cloudflare.com ↗
- Jared Kaplan and co-authors at OpenAI describe scaling laws: a language model's loss falls as a power law as the number of parameters, the amount of data and training compute growScience · Large language model · arxiv.org ↗
- F5 announces that it will buy Shape Security, which protects websites from bots with polymorphic code, for about $1 billionCybersecurity · Business & markets · Moving target defense · sec.gov ↗
- Microsoft invests $1 billion in OpenAIBusiness & markets · AI labs · globalventuring.com ↗
- The US adds Huawei and 68 of its foreign affiliates to the Entity ListGeopolitics & defense · Business & markets · Compute export controls · govinfo.gov ↗
- GPT-2: OpenAI initially withholds the full model over the risk of misuseScience · Media & information · Large language model · openai.com ↗
- AlphaStar beats professionals at StarCraft IIScience · Sport · Reinforcement learning · engadget.com ↗
- Google releases reCAPTCHA v3: instead of pictures, an invisible score from 0 to 1 based on how a person behaves on the websiteCybersecurity · Privacy & surveillance · Internet & infrastructure · Behavioral biometrics · security.googleblog.com ↗
- Google BERTScience · Large language model · arxiv.org ↗
- OpenAI GPT: a transformer pre-trained on textScience · Large language model · openai.com ↗
- The Cambridge Analytica scandal: data from tens of millions of Facebook profiles used for political advertisingElections & politics · Privacy & surveillance · Microtargeting · privacyinternational.org ↗
- MIT: early Parkinson's disease is detected from ordinary typing at home with an accuracy (AUC) of 0.76Health & medicine · Privacy & surveillance · Science · Behavioral biometrics · dspace.mit.edu ↗
- AlphaZero masters chess from scratch in hoursScience · Sport · Reinforcement learning · arxiv.org ↗
- The paper “Attention Is All You Need”: the Transformer architectureScience · Transformer · arxiv.org ↗
- Mastercard buys NuData Security, a “passive biometrics and behavioral analytics company”Finance & money · Cybersecurity · Business & markets · Behavioral biometrics · investor.mastercard.com ↗
- GitHub publishes a formal specification for GitHub Flavored Markdown, based on CommonMarkSoftware development · Markdown · github.blog ↗
- Google launches OSS-Fuzz, continuous fuzzing for open-source projects; by May 2025 it had helped find and fix more than 13,000 vulnerabilities and 50,000 other bugs in 1,000 projectsCybersecurity · Software development · AI vulnerability discovery · github.com ↗
- Final of the DARPA Cyber Grand Challenge: automated systems with no human involvement find, fix and exploit vulnerabilities; Mayhem from ForAllSecure winsCybersecurity · Software development · AI vulnerability discovery · darpa.mil ↗
- Amodei, Olah and co-authors from Google, Stanford, Berkeley and OpenAI publish “Concrete Problems in AI Safety”: side effects, reward hacking, scalable oversight, safe exploration and distributional shiftScience · AI safety · arxiv.org ↗
- AlphaGo beats Lee Sedol 4–1Science · Culture & arts · Reinforcement learning · koreatimes.co.kr ↗
- The Shuffler system (Columbia University) reshuffles the code of a running program every 50 ms at the cost of a 14.9% slowdownCybersecurity · Science · Moving target defense · cs.columbia.edu ↗
- OpenAI is foundedScience · Business & markets · AI labs · openai.com ↗
- MIT: ordinary typing shows that a person is sleep-deprived, with a detection accuracy (AUC) of 0.93Health & medicine · Privacy & surveillance · Science · Behavioral biometrics · pmc.ncbi.nlm.nih.gov ↗
- The US adds the Chinese military university NUDT and three national supercomputing centers to the Entity List: according to the authorities, the Tianhe supercomputers were believed to be used in nuclear explosive activitiesGeopolitics & defense · Compute export controls · govinfo.gov ↗
- Google reports that AI solves the hardest variant of distorted text with 99.8% accuracy, so such text is “no longer a dependable test”, and switches reCAPTCHA to the “I'm not a robot” checkbox with behaviour analysisCybersecurity · Internet & infrastructure · Death of CAPTCHA · security.googleblog.com ↗
- The first ACM academic workshop on moving target defense (MTD 2014) takes place in ScottsdaleCybersecurity · Science · Moving target defense · csis.gmu.edu ↗
- Jeff Atwood and co-authors release a unified “Standard Markdown” spec; after John Gruber objects, the project is renamed the next day and is now CommonMarkSoftware development · Markdown · blog.codinghorror.com ↗
- MIT Lincoln Laboratory researchers write that no memory randomization technique survives a leak of an arbitrary region of memory, and that 16 bits of randomness on an automatically restarted server are guessed within minutesCybersecurity · Science · Moving target defense · ll.mit.edu ↗
- Google buys DeepMindBusiness & markets · Deep learning · thedomains.com ↗
- Shape Security comes out of stealth with ShapeShifter, an appliance that rewrites a web page's code every time it is served, producing “an unending stream of randomized code variations” against botsCybersecurity · Internet & infrastructure · Moving target defense · siliconangle.com ↗
- Google's neural network for house numbers from Street View solves the hardest reCAPTCHA category with 99.8% accuracyCybersecurity · Science · Death of CAPTCHA · arxiv.org ↗
- DeepMind: a neural network learns to play Atari games from screen pixelsScience · Reinforcement learning · arxiv.org ↗
- Frey and Osborne (Oxford): about 47% of US employment is in occupations at high risk of computerisation; the higher the wages and education, the lower the riskWork & jobs · AI and jobs · oxfordmartin.ox.ac.uk ↗
- Touchalytics: the way a person scrolls a phone screen identifies them almost without errors within a session and with an error rate of up to 4% a week later; the authors see it as an addition to a password, not a replacementCybersecurity · Science · Behavioral biometrics · arxiv.org ↗
- Chu and co-authors: mouse movements and keystrokes identify 97.9% of blog bots with 0.2% false alarmsCybersecurity · Science · Behavioral biometrics · eecis.udel.edu ↗
- A Google neural network teaches itself to recognize cats in YouTube videosScience · Deep learning · 9to5google.com ↗
- The US federal cybersecurity research plan names the moving target one of four “game-changing” themes: systems must become “less deterministic, less homogeneous, and less static” in order to “substantially increase the cost of attacks”Cybersecurity · Laws & regulation · Science · Moving target defense · nitrd.gov ↗
- Zheng, Paloski and Wang: mouse movements identify a user with a 1.3% error rate after 20 clicks, but on a real forum 20 clicks took up to 38 minutesCybersecurity · Science · Behavioral biometrics · eecis.udel.edu ↗
- The collection “Moving Target Defense” comes out, edited by Sushil Jajodia and co-authors (Springer)Cybersecurity · Science · Moving target defense · doi.org ↗
- The Google Brain project startsScience · Deep learning · en.wikipedia.org ↗
- DeepMind is foundedScience · Business & markets · Deep learning · en.wikipedia.org ↗
- Motoyama et al. (University of California, San Diego): human CAPTCHA solving is sold from $1 per thousand; workers in Bangladesh, China, India and Vietnam are paid $0.5–0.75 per thousandWork & jobs · Cybersecurity · Business & markets · Death of CAPTCHA · usenix.org ↗
- Bursztein et al. (Stanford) have humans solve more than 318,000 CAPTCHAs of 21 types: three people give the same answer on only 71% of image CAPTCHAs and 31% of audio onesCybersecurity · Science · Death of CAPTCHA · web.stanford.edu ↗
- Google buys reCAPTCHA, which protects more than 100,000 websites from spam and fraudBusiness & markets · Cybersecurity · Death of CAPTCHA · googleblog.blogspot.com ↗
- Bigham and Cavender: 89 blind users solve 10 popular audio CAPTCHAs only 43% of the timeHealth & medicine · Internet & infrastructure · Science · Death of CAPTCHA · cs.cmu.edu ↗
- Killourhy and Maxion (CMU): 51 people type the same password 400 times; the best of 14 algorithms recognises a person by typing rhythm with an error rate of about 10%Cybersecurity · Science · Behavioral biometrics · cs.cmu.edu ↗
- Science publishes reCAPTCHA: while passing the check, people also recognise words from old books; in the first year more than 1.2 billion CAPTCHAs were solved and more than 440 million words decipheredInternet & infrastructure · Science · Culture & arts · Death of CAPTCHA · cs.cmu.edu ↗
- Bryan Ford and Jacob Strauss (MIT) propose “pseudonym parties”: people gather in one place at the same time and each receives one anonymous credential, so that “one real person gets one virtual persona”Privacy & surveillance · Science · Proof of personhood · bford.info ↗
- Microsoft announces that address space layout randomization (ASLR) is on by default in Windows Vista: a library is loaded into one of 256 locations, and an attacker guesses the address with a probability of 1/256Cybersecurity · Software development · Moving target defense · learn.microsoft.com ↗
- Shacham and co-authors show that memory randomization on a 32-bit system can be beaten by brute force: the Apache web server protected by PaX is broken in 216 seconds on averageCybersecurity · Science · Moving target defense · hovav.net ↗
- John Gruber introduces Markdown, an easy-to-read, easy-to-write plain text format for web writersSoftware development · Media & information · Markdown · daringfireball.net ↗
- Facebook launchesMedia & information · Mind & relationships · Social networks · historyhit.com ↗
- John Douceur (Microsoft Research) describes the “Sybil attack”: without a central authority vouching for identities, one participant can pose as manyCybersecurity · Science · Proof of personhood · microsoft.com ↗
- Net Nanny releases BioPassword, a Windows NT login that checks not only the password but also how the person types itCybersecurity · Software development · Behavioral biometrics · rcpmag.com ↗
- Luis von Ahn, Manuel Blum, Nicholas Hopper and John Langford of Carnegie Mellon University invent CAPTCHA, the “Completely Automated Public Turing test to tell Computers and Humans Apart”Cybersecurity · Internet & infrastructure · Proof of personhood · captcha.net ↗
- AltaVista employees file a patent application for a “human or program” check: to add a website to the search engine, you have to read a randomly distorted stringCybersecurity · Internet & infrastructure · Death of CAPTCHA · patents.google.com ↗
- The term “open source”, suggested by Christine Peterson, is coined at a meeting in Palo Alto; the Open Source Initiative is founded in late FebruarySoftware development · Business & markets · Open source · opensource.org ↗
- Ward Cunningham opens WikiWikiWeb, the first wiki; text there is marked up with ordinary characters: two apostrophes for italics, three for bold, an asterisk for a list itemSoftware development · Media & information · Markdown · wiki.c2.com ↗