OpenAI shipped GPT-5.6 (1,332 HN — top HN post) and named it the preferred model in Microsoft 365 Copilot, moving despite the ongoing MSFT-OpenAI breakup chatter. Fidji Simo — OpenAI's No. 2 and CEO of AGI Deployment — stepped down from her full-time role due to illness. Ars documented what may be a fatal OpenAI misstep in the NYT copyright fight: OpenAI is accused of faking inability to search training data and deleting billions of ChatGPT logs. Elon Musk publicly praised Mythos and Fable and pledged not to 'cut off' Anthropic — a striking reversal ahead of Anthropic's IPO. OpenAI shut down its Atlas browser after less than a year. Anthropic announced Claude Fable 5 will move to usage-based fees for consumers. TC notes Anthropic, OpenAI, and SpaceX together will generate more value than all US VC-backed exits since 2001. Meta's carbon emissions rose 25% last year.
OpenAI shipped GPT-5.6 and its new model family, and OpenAI simultaneously announced GPT-5.6 is the preferred model in Microsoft 365 Copilot — despite the ongoing breakup chatter. Fidji Simo stepped down as OpenAI's CEO of AGI Deployment after medical leave. OpenAI also announced ChatGPT Work and a Codex rebrand promising 'hours-long' agent runs.
GPT-5.6 ships (1,332 HN points — top HN post) — the model at the center of the two-week federal-vetting saga finally in production.
OpenAI: GPT-5.6 is Microsoft 365 Copilot's preferred model across Word, Excel, and Outlook.
TC: Fidji Simo, OpenAI's CEO of AGI Deployment, is stepping down from her full-time role after medical leave.
The Verge: OpenAI paired the GPT-5.6 rollout with a new 'ChatGPT Work' offering post-federal-greenlight.
Ars: The rebranded Codex promises independent workflows that can run 'for hours if needed.'
The NYT alleges OpenAI hid evidence in the ChatGPT copyright trial — specifically, faked inability to search training data and deleted billions of ChatGPT logs. Ars characterizes it as a potentially fatal misstep that could bring sanctions. Directly compounds the Cloudflare payment deadline and the Warren-Scanlon health-data bill.
TC: The NYT says OpenAI hid tools and datasets that could identify copyrighted journalism in the ChatGPT copyright trial.
Ars: OpenAI is accused of faking inability to search training data and hiding/deleting billions of ChatGPT logs — potential sanctions on the table.
Musk publicly praised Mythos and Fable and pledged not to cut off Anthropic — striking reversal ahead of Anthropic's IPO, worth ~$40B in revenue. Anthropic said Claude Fable 5 will move to usage-based fees for consumers. MIT Tech Review documented Anthropic's discovery of a hidden concept space inside Claude — the clearest interpretability glimpse yet.
TC: Musk publicly praised Mythos and Fable and pledged not to cut off Anthropic — with ~$40B in revenue at stake.
Wired: Anthropic will move Claude Fable 5 to usage-based fees for consumers — Anthropic's premium consumer tier gets pricier.
MIT TR: Anthropic developed a technique giving the clearest glimpse yet at a hidden concept space where Claude reasons.
The Verge: Anthropic launches Claude Wrapped — Spotify-style AI usage recap.
TC: Claude's Reflect dashboard reinforces AI habit — not just visualization.
TC: Anthropic, OpenAI, and SpaceX together will generate more value than every US VC-backed exit since 2001. Ollama raised $65M with 9M users. Paris-based voice startup Gradium raised $100M seed with Nvidia backing. Lyzr let its own agent run a $100M fundraise.
TC: The three AI IPOs will generate more value than all US VC-backed exits since 2001 combined.
TC: Ollama raised $65M with ~9M users, 176k GitHub stars.
TC: Paris voice startup Gradium raised $100M seed with Nvidia backing.
TC: Lyzr used its own AI agent to run a $100M enterprise-agent fundraise.
Meta entered the AI-coding fight with Muse Spark 1.1. OpenAI shut down its Atlas browser after less than a year. Meta's new AI chips will begin production in September. TC published a piece arguing Nvidia is a victim of the compute marketplace it created. Google will disclose AI-made ads.
TC: Meta's Muse Spark 1.1 pitches large agentic workloads, bug fixing, and coding performance.
TC: OpenAI sunsets Atlas after less than a year — but is moving its capabilities into ChatGPT itself.
TC: Meta's modular AI chips enter production in September.
TC argues Nvidia is now caught in the compute-marketplace-commoditization it enabled.
TC: Google will label AI-generated ads across Search, Discover, and YouTube.