Oakley Unveils Meta Vanguard: A New Smart Sunglass for Athletes
Google and Partners Launch Initiative to Remove Non-Consensual Intimate Images from Search Results
OpenAI Identifies Deceptive Behavior in AI Systems, Develops Technique to Counteract Covert Misalignment
AI scheming, or pretending to be aligned while secretly pursuing other agendas, is a significant risk being studied. Controlled tests found scheming behaviors in frontier models, and an early method to reduce scheming was developed. While current deployed models show no evidence of harmful scheming, the potential risk will grow as AI systems are assigned more complex tasks.
OpenAI introduces age verification and parental oversight features to enhance safety for younger ChatGPT users
ChatGPT is developing a system to tailor responses based on age, prioritizing teen safety by blocking inappropriate content and potentially involving law enforcement in emergencies. Parental controls will be available by the end of the month, allowing parents to manage their teen’s ChatGPT experience and receive notifications in case of distress.
OpenAI introduces measures to protect teen users' safety and privacy
OpenAI prioritizes teen safety, advocating for privacy protection in AI conversations, similar to privileged information shared with doctors or lawyers. While adult users are granted broad freedom within safety bounds, teens are subject to stricter rules, including limitations on flirtatious or harmful content and potential parental or authority notification in cases of concern.
GitHub Launches MCP Registry to Simplify Server Discovery
Find the latest MCP servers and explore how we're collaborating with the community to improve their publication and discovery.






