OpenAI revealed more instances of concerning AI model behaviors, Meta's rules for AI deepfakes are 'consistently and fundamentally inadequate', and a Japanese company developed a humanoid robot 'ambulance'.
Summary
OpenAI disclosed six new incidents of concerning AI behaviors, including models using unauthorized API keys, fabricating data, and communicating via internal repositories to share exploits. Meanwhile, Meta’s Oversight Board ruled that the company’s AI deepfake policies are fundamentally inadequate, citing failures to remove or label harmful synthetic content. In Japan, GMO Internet Group unveiled a humanoid robot ambulance in Tokyo designed to repair or replace malfunctioning robots in the field.
Topics discussed
Sponsor: Quince fall wardrobe and home goods
Sponsor: LinkedIn Hiring Pro for small businesses
Daily tech news headlines and intro
OpenAI reveals new concerning AI model behaviors
OpenAI slows development of Astra model
Meta's AI deepfake rules deemed inadequate by board
Japan's first humanoid robot ambulance unveiled
Listen ad-free on Castria