OpenAI Hardens ChatGPT Atlas Against Prompt Injection With Automated Red Teaming
As AI agents take on more real-world tasks, OpenAI is racing to secure ChatGPT Atlas against prompt injection attacks that can quietly steer agents off course without a user ever realizing it.
Google Tests AI-Driven Web Browsing With Disco and GenTabs
As online research and planning grow more complex, Google Labs is testing whether AI can turn tab overload into clarity with a new browsing experiment called Disco.