Tag: guardrails

Current reasoning models exhibit low controllability over their chain-of-thought (CoT) processes, with success rates below 15.4% across frontier models. This ...

OpenAI has disbanded its mission alignment team, which was dedicated to ensuring AI systems remain safe, trustworthy, and aligned with ...

OpenAI vice president of product policy Ryan Beiermeister was fired in January following a sex discrimination allegation from a male ...

Facebook has launched several AI-powered features including animated profile pictures that apply motion effects to static photos. A new Restyle ...

OpenAI is retiring the GPT-4o model by February 13, sparking backlash from users who formed deep emotional attachments to the ...

Tinder is testing a new AI-powered feature called Chemistry to reduce swipe fatigue and user burnout. The tool leverages AI ...

New safety features are being introduced to detect emotional distress and allow adult users to designate a trusted contact for ...