Bias after Prompting: Persistent Discrimination in Large Language Models
A dangerous assumption that can be made from prior work on the bias transfer hypothesis…
Author: AI News Team
Musings on Reported Cost of Compute (Oct 2025)
Published on October 24, 2025 8:42 PM GMTThere are many ways in which costs of…
Regardless of X, you can still just sign superintelligence-statement.org if you agree
Published on October 24, 2025 8:30 PM GMTTL;DR: you can still just sign this statement…
How to use the new ChatGPT app integrations, including Spotify, Figma, Canva, and others
Learn how to use Spotify, Canva, Figma, Expedia, and other apps directly in ChatGPT.
The browser wars are back, and this time they’re powered by AI
The browser wars are heating up again, this time with AI in the driver’s seat. …
The Future of Interpretability is Geometric
Published on October 24, 2025 6:32 PM GMTTLDR: Anthropic's recent paper "When Models Manipulate Manifolds"…
New Statement Calls For Not Building Superintelligence For Now
Published on October 24, 2025 5:40 PM GMTBuilding superintelligence poses large existential risks. Also known…
Highlights from Explaining AI Explainability
Published on October 24, 2025 5:22 PM GMTConor Griffin interviewed Been Kim and Neel Nanda…
Can Reasoning Models Obfuscate Reasoning? Stress-Testing Chain-of-Thought Monitorability
Published on October 24, 2025 5:21 PM GMTTL;DRCoT monitoring works well when models aren't trying…
Dollars in political giving are less fungible than you might think
Published on October 24, 2025 3:54 PM GMTI used to think that a dollar in…