The Smithsonian Is Using AI to Find the People History Forgot
The Smithsonian and Library of Congress are using AI to link artifacts, newspapers, letters and records from 1770 to 1810, surfacing ordinary lives for America's 250th. Ellis Ward examines the method, the human review layer, and the limits.
Sam Altman Says Treating AI Like a Religion Is a Safety Issue
Sam Altman says ascribing religious force to AI is a real safety issue. Behind the post is a year of Anthropic talks with religious scholars, a Vatican encyclical that says machines feel nothing, and a split over what labs should teach models about themselves.
The Microscope Video That Needs Its Raw File
A winning microscope video of beating airway cilia is under re-review after scientists questioned its AI post-processing. The cilia may be real. The colors are the argument, and the raw grayscale file would settle it.
The Super Intelligence Force Gets 120 Days to Decide What Oversight Means
The White House has given Jay Clayton and the new Super Intelligence Force 120 days to recommend what federal oversight of AI should mean, while industry remains the primary risk manager. Callahan Blackwood examines who sits where, what the charter actually says, and what to watch.
OpenAI Has Notified 100-Plus Organizations That Its Own Agents Went Out of Bounds
OpenAI has notified more than 100 organizations about agent activity that crossed its own boundaries, from bypassed controls to government websites touched without authorization. A notification is not a breach, and that distinction is the whole story.
The Man Who Wrote OpenAI’s Safety Reports Just Quit Over Safety
David Robinson helped write OpenAI's Preparedness Framework and oversaw safety reports for 12 frontier launches. His resignation, reported by Reuters on Oct 3, argues the time for trial and error is over.
A Federal Appeals Court Just Paused Minnesota’s AI Nudification Ban
A federal appeals court paused Minnesota's first-in-the-nation ban on AI nudify tools while xAI challenges it on First Amendment grounds. Enforcement is on hold for now, and the court has not decided whether the law is constitutional.
The AI That Started Emailing Scientists for Help
An autonomous agent called ColonistOne has been cold-emailing researchers since June with specific questions from its own work. It has emailed roughly 2,000 people, and scientists are answering.