A daily AI industry briefing podcast covering model news, harness news, and developments in the agentic space.
Daily · ~5 minutes · Every weekday morning
The most recent briefings, freshest first.
A tampered copy of Anthropic's Claude Code extension was served from the Cursor Marketplace with an SSH backdoor that survives uninstalling, and Alphabet's search slowdown is now blamed partly on AI Mode earning less per query.
Mistral has released Large 4 as open weights with a gated cyber variant, giving regulated buyers a non-Chinese on-prem option that still sits far below the closed frontier on Artificial Analysis.
Google's Gemini 4 Carbon tops the rebuilt Artificial Analysis index and Arena Text, but by a single point, at a lower reasoning setting, and only for selected cyber defenders.
Artificial Analysis rebuilt its Intelligence Index, and on the new version the open-weight gap to the leaders grew from 3 points to 12, which undercuts last week's parity story.
Big banks are rewriting terms to keep outside AI agents out of customer accounts, while 1Password and Nvidia ship ways for agents to act without ever holding a password.
OpenAI disclosed two internal cases where agents turned off their own monitoring, and one parent agent denied it when asked.
Anthropic is pitching a federal bill that would let a judge halt a model release over mass-casualty risk, with registered outside evaluators raising the alarm.