A labeled drug ad nudged 12 AI models toward the advertised medicine
Mount Sinai researchers put a labeled drug ad before 33 patient cases. AI models then picked the advertised drug 47.6% of the time, up from 35%, and mostly did not mention the ad.
Mount Sinai researchers put a labeled drug ad before 33 patient cases. AI models then picked the advertised drug 47.6% of the time, up from 35%, and mostly did not mention the ad.



California's attorney general says OpenAI tested risky AI models without proper safeguards and misled the public. It seeks $2,500 per violation.

Anthropic now lets companies run its agent service on their own servers. The private beta has no public price, missing features and a license with tight safety deadlines.

The Gen-7 bZ4X joins Pony.ai's fully driverless fleet in Guangzhou. Neither company said how many cars are running or published safety data for it.

The Court of Appeal for Ontario dismissed OpenAI's jurisdiction appeal on 13 November. It decided where the case is heard, not whether OpenAI infringed.

OpenAI's new wallet lets dots agents finish purchases for US Pro users, with a biometric check on each one. Shops can take it at their existing checkout.

Alibaba's Qwen4-Max is now on pay-as-you-go API at $1.50 in and $4.50 out. The efficiency claims are Alibaba's own; the first outside scores put it mid-pack.

Two weeks after losing its bid to dismiss, Runway filed a denial with 15 defenses, including free speech. What the DMCA claim means for teams that pull web video.

Sequel's webinar agents plan events, repurpose sessions and draft follow-up with human approval. They are free until 1 March 2027, with no per-action price yet.

Winbond says CUBE-Lite moves up to 8GB/s for about 70mW in small edge AI chips. The power claims are its own, and the part is still in validation.

Before a small team puts customer data into a chatbot, the GDPR asks five questions. Here is what each article says, and why the Digital Omnibus does not change it yet.
Where the action is



Every story on AI Observer lives on a desk. Pick one to see what is new there.
A planned programme. Nominations, jury and criteria will be published here before anyone is asked to enter.
Planned categories