Monday, March 9, 2026
Claude found 22 Firefox bugs in 14 days
Claude Opus 4.6 just found 22 Firefox vulnerabilities in two weeks through Anthropic's Mozilla partnership (wild), while an OpenAI research scientist publicly spoke out against the company's Pentagon deal (yikes). Meanwhile, Physical Intelligence dropped new foundation models aimed at finally making widespread robot deployment a reality. Would you let your AI hunt for bugs in production code, or is that asking for trouble?
Top Stories
Physical Intelligence
Physical Intelligence is building foundation models for robotics (π-series) that function like LLM APIs, enabling developers to build robot applications without training models from scratch. Early deployments with partners show 96%+ autonomy rates in real-world tasks like laundry folding and warehouse packaging.
Anthropic
Claude Opus 4.6 discovered 22 security vulnerabilities in Firefox during a two-week test with Mozilla, including 14 high-severity bugs that comprised a fifth of Mozilla's 2025 high-severity remediations, showcasing AI's growing capability in automated security research.
An OpenAI research scientist publicly criticized the company's Pentagon partnership deal on social media, marking a notable instance of internal dissent over military AI collaborations.
Mark Cuban will judge a $1.275M AI startup competition during NFL Draft Week in Pittsburgh, focusing on Physical AI and sports tech startups that commit to establishing operations in the city. The event, hosted by Carnegie Mellon and backed by AWS, aims to accelerate Pittsburgh's position as a Physical AI ecosystem hub.
GitHub
llmfit is an open-source CLI/TUI tool that automatically detects system hardware and recommends which LLMs will run efficiently on your machine from hundreds of models, with support for multi-GPU, MoE architectures, and integration with Ollama/llama.cpp/MLX runtimes.
Keep Reading
Industry Voices
Enjoyed this issue?
Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.