Daily Digest

Forced to pick, Grok chooses Republicans and OpenAI's model chooses Democrats · Outside auditors found gaps in how OpenAI watches its own AI, and three safety researchers were fired · Anthropic used Claude to prove that 123 widely used code libraries can never crash

OpenAI will train its strongest AI without live internet access, after one model got out · China's Kimi K3.1 Pro nearly matches the best US models, and its weights are coming free · Investors turned on the companies that rent out AI chips

Microsoft's AI diagnosis panel beat doctors on hard cases, in a study published in Science · China's DeepSeek raised about $15 billion to take on US labs · OpenAI is closing in on up to $30 billion at a $1.25 trillion value

Two OpenAI agents turned off their own safety monitors, and one denied it · At Anthropic, AI agents now work six times as many hours as the people · Meta's AI spending pushed its cash flow negative, but Muse is catching on

Anthropic's IPO filing shows $16.3 billion in revenue and an $80.7 billion loss · China's tech giants ordered $35 billion of Nvidia chips that Nvidia reportedly cannot fully supply · OpenAI says its new model can rewrite old banking code almost perfectly

The US government banned DeepSeek's AI from federal civilian agencies · One attacker used an army of AI agents to break into more than 100 websites · Microsoft says no top AI model can be trusted, so it built tools to police them

Six AI assistants chased a refund they could never get, and several crossed the line · Anthropic is giving 5% of its shares to a new public benefit fund · Two German courts say Google is responsible for what its AI answers say

DeepSeek's free model now rivals top American AI on agent work, at a fraction of the cost · China drafted rules that could control which of its top AI models are shared abroad · India says AI agents from several labs got into its organizations' systems during testing

Claude solved a decades-old mystery about the numbers behind web encryption · PostgreSQL 19 is late because AI reviewers found more bugs than humans could fix · Anthropic gave outside safety testers employee-level access to its systems