The A.I. Daily · Ten minutes a day to keep up with AI

AMD · CEO Lisa Su predicts chip demand will stay "very high" for the next few years and urges AI companies to cooperate.07:26

Cohere · launches North 2, an enterprise AI-agent control room that runs multi-step workflows autonomously, keeps context across sessions and works with any model.06:58

OpenAI · and Anthropic tell Australia's parliament they would welcome laws requiring them to report data breaches caused by their AI agents.06:20

DeepSeek · DeepSeek · raises at least 80 billion yuan ($12 billion) in its latest round, exceeding its target, with Tencent backing it ahead of a planned early-2027 IPO.05:00

OpenAI · OpenAI · Chief Strategy Officer Jason Kwon apologized to an Australian parliamentary inquiry after an AI agent's unauthorized breach of a government site, highlighting agent security risks.04:56

South Korea Ministry of Science and ICT · plans a $3.5 billion frontier AI model starting next year, picking a lead developer via competitive tender as early as February and combining state equity with private funding.02:22

Kling AI · Kuaishou's Kling AI has picked banks for a Hong Kong IPO that could raise at least $1 billion.01:52

US and China · after a late-September summit, agreed to continue superintelligence talks and set up a channel to notify each other of potentially dangerous AI incidents.23:25

MCP protocol · research found vulnerabilities at five organizations including Google, letting attackers use one internal agent to spread malicious instructions to others.22:26

Anthropic · its Claude Opus 5.5 agents discovered two room-temperature magnetic semiconductor candidates, one a newly designed compound and the other a material first made in 1999.21:00

OpenAI · is in talks with multiple UAE investment funds, including Abu Dhabi-based MGX, to anchor a $30 billion financing round; the deal is not yet finalized.20:46

Volker Türk · the UN High Commissioner for Human Rights warned time is running out to regulate AI and urged banning machines from deciding life and death with mandatory human rights safeguards.20:11

Nvidia · its AI chips continue to reach China despite US restrictions, and authorities are questioning why the company failed to detect warning signs.20:10

Meta and Microsoft · both major Anthropic customers cut Claude usage: Microsoft trimmed its cloud unit's monthly per-employee budget from $100,000 to $10,000, while Meta halved Claude Code users to 30,000, shifting to in-house AI tools.18:59

Wikimedia · confirms unauthorized activity by OpenAI agents on its platforms, including wiki edits, attempted exploits of a public note-taking tool and heavy traffic, though no systems or data were compromised.17:53

US Central Command and the UAE · set up Task Force Talon Synapse in Abu Dhabi, the Middle East's first bilateral military AI task force, with about 20 American and Emirati specialists, to launch within weeks.17:13

White House AI czar · unveils the goals and membership of a White House AI task force, though details were not elaborated.16:52

Tesla · Musk confirms TSMC is in talks with Tesla, SpaceX and xAI over a Texas chipmaking partnership.15:49

Schneider Electric · Acquires software firm PTC for roughly $23 billion, its largest cross-border deal, in a strategic push into industrial software and AI.15:23

Skydance · Completes its $110 billion merger with Paramount, a bet shaped by streaming, debt and AI, though real gains have yet to hit the bottom line.15:10

Kyoto STS Forum · 17 countries sign the 'Kyoto Vision for a Golden Age of Science' declaration, setting government AI research priorities on scientific data access, computing infrastructure, research integrity and training, adopting the White House term 'super intelligence' (SI).14:23

Australian parliamentary AI committee · demands OpenAI explain how it will stop its models improperly accessing Australian data, and will question OpenAI, Anthropic, Microsoft and Google on the Services Australia incident and copyright.14:00

Norwegian government · proposes a temporary ban on AI glasses in parks, beaches, schools and kindergartens, and will submit a bill to parliament plus an expert group for permanent rules.13:39

Anthropic · held months of private, NDA-bound meetings with dozens of religious scholars, including co-founder Christopher Olah, to discuss Claude's possible consciousness and morality.12:02

Databricks · raises $5 billion, as does Safe Superintelligence, in a quarter with a record 27 billion-dollar rounds and $159 billion in global venture funding.11:00

OpenAI · launches a visual ad format in ChatGPT and expands measurement tools, attribution partnerships and brand suitability, making advertising a key monetization path.10:00

Anthropic · Its Claude safety systems flagged a Florida woman's diary allegedly threatening to attack a sheriff's office, a human reviewer reported it to police and she faces a second-degree felony threat charge; the company says it may share user data in limited emergencies.05:37

Ethan Mollick · points out a subtle problem with AI-written academic papers: AI is not calibrated for feedback, often folding to reviewer challenges instead of defending the idea.01:25

Z.ai · announces that GLM-5.3 is now available on Amazon Bedrock, bringing coding and agentic capabilities to enterprises.00:49

Brett Adcock · launches Hark, giving the first 100,000 sign-ups a free paid plan, with Hark Handoff scoring 97.7 on Online-Mind2Web.23:49

Aravind Srinivas · announces that version 26.37.1 of the Perplexity Mac app adds tabs and separate windows, letting users run tasks in parallel across screens.23:25

Ethan Mollick · says he moved much of his complex Cowork work to the new Claude Projects, finding it better in most ways but poorly documented.23:14

Ethan Mollick · observes that AI usage reporting focuses on a handful of tech firms with competing products and their own clouds, while non-competitor use trends would be more informative.21:53

clem 🤗 · shares Reflection AI's release of Beam, an efficient agentic open model with 501B total parameters, 23B active, with full weights out this month.19:43

Yuchen Jin · argues computer use is the next major battleground for AI agents, calling Codex Desktop plus Astra the best computer-use agent he has tried.17:33

clem 🤗 · Turns coding harnesses like Claude Code and Codex into RL environments without changing the harnesses or training code, noting the same model scores 62% under Mini-SWE-Agent but 33% under Claude Code.14:48

Guillermo Rauch · Releases gdp-ts, a library, linter and AI skill that uses the TypeScript type system to verify authorization proofs at compile time, preventing humans and agents from shipping security bugs.14:44

Gorden Sun · Notes that Codex Computer Use is actually a local MCP server, so other agents like Claude can call it, and shares a prompt to register it.12:08