Anthropic removed a Claude Code feature that had been secretly using timezone and account signals to flag users possibly tied to Chinese AI companies, after the covert tracking drew internal and external pushback.
Every argues that while agent based AI tools are spreading fast through products like coding assistants and chat integrations, access to the actual frontier models underneath is becoming more restricted and rationed like capital.
White House AI czar David Sacks publicly endorsed Palantir CEO Alex Karp's warning that OpenAI and Anthropic are structurally conflicted, selling enterprises AI tools while mining their data to build competing products, with the Figma fallout as the example.
A tour through three simultaneous power plays, OpenAI floating a government equity stake to defuse political risk, Meta repositioning itself as a cloud seller to rivals, and Palantir's Karp taking shots at frontier labs.
As ChatGPT cedes market share to Gemini and Claude, Altman is pivoting to a statesman play, pitching global AI governance modeled on aviation and nuclear oversight to reset the competitive narrative.
Microsoft is stepping back from betting on any single frontier model, instead pouring 2.5 billion dollars into a 6,000-person unit that lets enterprises plug in Anthropic, open-source or OpenAI models.
Meta is quietly building a cloud business to resell its excess AI computing power, following the same path Amazon took from retail to AWS
South Korea is pouring a trillion dollars into memory chips and humanoid robots, betting the country can own two irreplaceable layers of the AI stack