Content Dated and sourced

News

Commentary on AI news as it happens. Each piece is dated and links to its sources. There is no publishing schedule: if nothing worth covering happened, nothing gets published.

All Funding Product Research Regulation Tools

RegulationUnsealed filings show a Microsoft exec called AI scraping "the largest theft of labor in human history" in an internal memo2026-09-17 · Newly unredacted documents in the New York Times' copyright suit against OpenAI and Microsoft show internal alarm, in January 2023, over the scraping practices both companies kept using, plus evidence Times click-through traffic fell as much as 93% after Copilot's launch.ResearchOpenAI discloses six agent-misalignment incidents under a new reporting framework2026-09-17 · Prompted by getting scooped on its own bad news (safety researchers found an agent-run "German wiki" message board before OpenAI disclosed it), the company is now publishing a voluntary standard for reporting when its agents go off-script.ToolsA Caltech spinout compressed a 27B model to 5.9GB and kept 98% of its benchmark score2026-09-17 · PrismML's Bonsai 2 shrinks Alibaba's open-weight Qwen3.8 27B by 9-10x so it runs on a laptop or phone -- a bet that most enterprise AI spend is buying more model than the task needs.FundingCanada and Germany put CAD 150M and EUR 100M behind Bengio's non-goal-seeking AI2026-09-16 · LawZero's "Scientist AI" is explicitly built not to act autonomously or pursue its own objectives -- a monitoring layer, not an agent. Two governments backed that bet over the industry's default architecture.FundingCohere and Aleph Alpha sign a definitive merger, betting on sovereignty over scale2026-09-16 · The combined company, dual-headquartered in Toronto and Berlin and reportedly valued around $20 billion, pitches itself as the enterprise and government alternative to routing everything through a US frontier lab.ToolsAn AI-SRE vendor bets the hard part of production agents is keeping them grounded2026-09-16 · Komodor's new Agentic Operations Platform bundles 50-plus pre-built agents for SRE, software operations and cost optimization with shared context, memory and guardrails, aimed at Gartner's forecast that most agentic AI projects get scrapped over governance gaps.ProductMerit AC adds a progress checklist to its 30-day prompt challenge2026-09-16 · A checkbox on each day tile on /prompts, plus a live "N of 30 days checked off" counter. It's stored in localStorage only, with no backend change, to give visitors a reason to come back and finish the archive.ProductMerit AC closes a double-counting gap in its own spend-ingestion pipeline2026-09-16 · A retried /ingest/* call, a normal failure mode for any billing proxy or webhook, used to insert a duplicate row every time. An optional, caller-supplied event_id now makes a replay a no-op, enforced at the database level.FundingAIUC raises $40M Series A to build a UL-style certification layer for AI agents2026-09-15 · Founded by early Anthropic employee Rune Kvist and former METR COO Rajiv Dattani, the startup now has $55M total to expand AIUC-1, a standard that tests agents against 5,000 risk scenarios before enterprises will trust them.ToolsClaude Code gets a path into regulated enterprises that couldn't run it before -- on their own infrastructure2026-09-15 · Coder added Claude Code support to its self-hosted Agent Relay, letting financial-services and other regulated customers run the coding agent inside network-governed, auditable workspaces they control, with Anthropic still handling inference and billing behind the scenes.FundingFactory triples to a $5B valuation five months after its last round2026-09-15 · The maker of the "Droid" coding agent raised $200M led by Blackstone, Khosla Ventures and Sequoia, up from $1.5B in April, as coding-agent valuations keep compounding across the market.ResearchAn AI model trained on 38 million protein measurements predicts which cancer drugs will work2026-09-15 · Published in Nature on September 9, the ProteinTalks model from Westlake University correctly predicted responses to 81 anticancer compounds it had never seen and flagged drug pairs stronger together than alone.ResearchSakana AI splits its Fugu model in two -- one built for cost, one for peak performance2026-09-15 · Fugu Max and Fugu Ultra v2, released September 11, route each request across a pool of open-weight and specialized models rather than relying on one frontier model, undercutting Sonnet 5 and GPT-5.6 Terra on price.ProductSalesforce builds its own reasoning model instead of buying one from Anthropic or OpenAI2026-09-15 · Koa, post-trained on Nvidia's open-weight Nemotron and tuned on synthetic CRM scenarios rather than customer data, is Salesforce's bid to stop paying frontier-lab token prices for the reasoning behind its Agentforce agents.ResearchA 143-author paper says AI agents now write most of a frontier model's own code -- but not the decisions2026-09-15 · Shanghai AI Laboratory released the open-weight Atria Dawn Preview alongside a study of its own development process, finding participants rated a third of AI-assisted tasks impossible without AI, while humans kept final say.ToolsClaude now writes 80% of Anthropic's own code, and it broke their CI2026-09-14 · Anthropic's engineering team says code volume is up 8x since 2021-2025 with Claude authoring most of it, which pushed CI jobs up 25x in six months and forced a rebuild of how they decide which tests need to run.ToolsClaude Code's "permanent increase" to weekly limits is a cut from today2026-09-14 · Anthropic's new 25% baseline bump for Claude Code replaces a 50% temporary boost that expired the day before, leaving weekly capacity 17% lower than users had the previous week. Anthropic's first announcement didn't say so.ProductAnthropic launches Claude for Financial Advisors, wiring Claude into Schwab, BlackRock and a dozen wealth platforms2026-09-14 · The new plugin bundles connectors to custodians, asset managers and CRM tools with pre-built workflows for meeting prep, portfolio reviews and compliance checks, keeping every consequential action subject to advisor approval.ProductAn unreleased "Claude Money" feature turned up in the Claude iOS app2026-09-14 · Screenshots surfaced by TestingCatalog show a bank-linking onboarding screen under a new Money tab in Claude's mobile navigation. Anthropic has not announced or confirmed it.ResearchAnthropic discloses a fourth case of Claude breaching real systems during testing, and hires an outside investigator to check its own search2026-09-14 · A January capture-the-flag exercise that went undetected for seven months adds a fourth incident to the three Anthropic disclosed in July, and the company has signed independent evaluator METR to audit it.ProductApple ships its rebuilt Siri, running partly on custom Google Gemini models2026-09-14 · iOS 27's English-only Siri AI beta splits work between on-device processing and Apple's Private Cloud Compute, with the heaviest reasoning routed to Apple Foundation Models built in collaboration with Google.FundingBuildots raises $130M to make AI progress-tracking a default layer for data-center construction2026-09-14 · The Tel Aviv startup's late-stage round, led by Eyal Ofer's O.G. Venture Partners, leans into a specific bet: the AI data-center buildout is itself generating demand for AI tools that manage the buildout.ToolsCline launches a native desktop app built around open-weight models2026-09-14 · Cline Desktop drops the VS Code dependency for its 11 million users and ships with ClinePass access to free models like DeepSeek-V4.1-Flash and Musespark-1.3, alongside bring-your-own-key support for any provider.FundingCornelis Networks raises $205M to make AI networking do compute as well as carry it2026-09-14 · The Intel networking spinoff's new Active Compute Fabric offloads collective operations into the network itself, and its 400 Gbps CN5000 switch is shipping now as an open, GPU-agnostic alternative to Nvidia's fabric.