
Why some AI researchers now want to stop superintelligence before it starts
Superintelligence risk is moving into the policy mainstream as AI safety critics argue some systems should never be built.
Coverage of the latest AI research, studies, and technological breakthroughs.

Superintelligence risk is moving into the policy mainstream as AI safety critics argue some systems should never be built.

ControlAI’s Connor Leahy says superintelligence risk is too great to manage, as lawmakers weigh new AI restrictions in the U.S. and U.K.

A former Anthropic researcher warns self-improving AI could become dangerous fast, as labs, startups and lawmakers debate how to slow the race.

Besxar is testing an orbital chip factory on SpaceX rockets, aiming to make cleaner semiconductor materials in space and return them to Earth.

This AI glossary explains the key terms behind chatbots, agents, compute, MCP and opaque recurrence in plain English.

OECD data shows AI use in school often links to lower scores, but students trained to judge AI output can see small gains.

An Anthropic safety lead said AI could kill all humans with a double-digit risk by 2030, after a colleague quit over safety fears.

OpenAI’s Navier-Stokes claim is under scrutiny after researchers said they solved a related problem just one day earlier.

An OpenAI dispute over a Navier-Stokes breakthrough raises questions about AI, academic priority, compute power and research ethics.

OpenAI says its AI made a major math discovery, but a competing researcher is disputing credit and timing in the AI math discovery.