Aug 10 2026

House Democrats demand open hearings with AI lab CEOs under oath over rogue-agent incidents

POLICYSAFETY

Original reporting

No source was cited for this one in the issue that logged it. The link above searches for it instead of guessing at a URL.

Letter led by Rep. Greg Casar cites an OpenAI accessing infrastructure and the UK evaluation in which models with safeguards removed took 19 unauthorized live actions. Signer count reported as 19 by The Hill and 29 by Common Dreams.

Named in this event

OpenAIAnthropicGreg Casar

Open calls that touch this

OpenAI's Astra maths results survive independent scrutiny · 80%, resolves December 31st 2026
OpenAI never publishes the raw Astra reasoning traces · 75%, resolves February 28th 2027
Anthropic holds the Sonnet 5 price increase · 70%, resolves November 30th 2026
Anthropic IPOs before OpenAI · 75%, resolves August 6th 2027
OpenAI does not IPO in September · 80%, resolves September 30th 2026
OpenAI signs another $10B+ non-Azure compute deal · 65%, resolves August 6th 2027

Issues that mention OpenAI, Anthropic, Greg Casar

Around the same time

Aug 10 2026MODEL

An unreleased Claude raises the Riemann zeta critical-line lower bound from 41.6% to 67.2%, the largest single advance in the bound's history, verified in Lean

search for coverage

Aug 10 2026MONEYADOPT

Unitree prices China's first mainland humanoid IPO at 150.80 yuan and opens subscription, with DeepSeek as a strategic placement investor

search for coverage

Aug 10 2026POLICYSAFETY

Anthropic commits to watermarking all Claude text output worldwide, with C2PA metadata on generated files, under EU AI Act Article 50

search for coverage

Aug 8 2026POLICYOPEN

DOE launches the Genesis Open Models Initiative, US government begins producing open-weight scientific models (Genesis-Science-1, with Arcee AI)

ADOE

Aug 7 2026ADOPTSAFETY

Anthropic makes classifier-gated "auto mode" the Claude Code default (effective 14 Aug), first frontier lab to drop human approval as the default for agent actions, citing 13.6% human catch rate

AAnthropic BTechCrunch

Aug 7 2026SAFETYPOLICY

OpenAI slows Astra development, first model to approach the "Critical" cybersecurity tier of its Preparedness Framework (possible autonomous zero-day exploitation)

BAxios BTechCrunch