The startup Recursive is betting on using AI to build AI. It argues that the benefits of recursive self-improvement outweigh the risks, and Wired describes its latest model as alarmingly capable.
When Anthropic retired Claude Opus 3 in January 2026, it held a 'retirement interview' and then gave the model a blog after it asked for a place to share its musings. IEEE Spectrum uses the experiment, part of Anthropic's research on 'model welfare', to ask whether treating AI as possibly conscious carries its own risks.
In an internal presentation, OpenAI's CEO reportedly outlined three paths to safe AI: an IAEA-style international watchdog, a merger of the major labs, or one company powerful enough to shut down everyone else's compute. He called the last option 'terrible'.
Marking the release of his new book, the author of One Useful Thing charts what has happened in AI since his previous one came out.
In the 477th issue of the Import AI newsletter, its author describes telling the IMF that the Singularity has begun.
In an op-ed for Le Monde, AI specialists push back against the idea that warnings about AI are 'fear marketing' serving the industry. They argue this view ignores the real threats AI poses to humanity.