AI Is Eating Itself. The Models Training on AI Content Are Getting Dumber.
By Chris Meredith | @ChristopherMeredith I’ve been watching a slow-motion catastrophe unfold for the past two years, and I’m not sure the people causing…
Read moreArchive
By Chris Meredith | @ChristopherMeredith I’ve been watching a slow-motion catastrophe unfold for the past two years, and I’m not sure the people causing…
Read moreBy Chris Meredith (@ChristopherMeredith) Tags: Technology, Artificial Intelligence, AI Safety, Machine Learning, OpenAI When OpenAI engineers found the message board in early July, they…
Read moreSubstack/Medium draft — v3 Status: v3 — date corrected (mid-2026), double-hyphens replaced throughout, report periods cited explicitly There’s a story that gets told about…
Read moreSubstack/Medium draft — v3 Status: v3 — DeepSeek-first recommendation (follows corrected numbers); double-hyphens replaced; URL corrected; date corrected The conversation around AI coding agents…
Read moreA SaferAI evaluation found that GLM-5.2, an open-weight model anyone can download, is months behind GPT-5.5 on capability but refused none of the dangerous…
Read moreGoogle DeepMind released Gemini Robotics 2, the first model to coordinate a robot's full body as a single system. Here is what changed, what…
Read moreIn February 2026, Anthropic published a report that reads like a corporate espionage thriller. A company called Moonshot AI had allegedly run one of…
Read moreThe numbers from Kimi K3’s release are real and they are significant. Ninety-one percent on BrowseComp, which tests whether a model can complete multi-step…
Read moreIn June 2026, researchers and engineers released thirteen new foundation models for embodied AI and humanoid robots. Thirteen in a single calendar month. That…
Read more15,000 people is not a rounding error. That is the rough total of Meta employees either laid off or reassigned away from their existing…
Read more