AI & Models
The latest from our AI & Models desk.
I can't tell if this design is prompt-engineered or genuinely good
A graphic design award shortlisted a campaign that was entirely AI-generated. The judges didn't know. The studio didn't disclose it. The debate over whether that matters is the most honest conversation the industry has had in years.
The AI radiologist found everything the human missed. Then the lawsuit started.
A London hospital deployed a chest X-ray model that caught 23% more abnormalities than its senior radiologists. Then a patient sued both the machine and the doctor who overruled it.
Open weights just learned to reason — and the gap moved again
A fully open model now walks through multi-step problems out loud. The capability that was supposed to be the closed labs' last fortress just opened.
Your new coworker is an AI agent. No one asked if you wanted one.
Enterprise rollouts of coding and operations agents tripled this year. The awkward secret: no manager has figured out how to performance-review a language model.
The agent that books your flights is quietly better than you
It prices fares at 3am, rebooks the cancellation before you wake up, and never argues with the gate agent. The human travel planner is having an identity crisis.
Everyone wrote AI safety laws. No one is enforcing them.
The EU AI Act, the U.S. Executive Order, the G7 code of conduct — all on the books. The enforcement bodies are understaffed, underfunded, and in one case, still looking for office space.
Open models just caught up to the closed giants — here is the data
On the MMLU-Pro and SWE-bench 2.0 leaderboards, the top open-weight model now trails the leading closed API by less than 1.5%. What that means for the labs spending billions to stay ahead.