Claude Security drops + Figure ships 350 robots

Claude

Anthropic's Claude Security is now in public beta for Enterprise customers, using Claude Opus 4.7 to scan code for vulnerabilities and generate fixes, with integration across major cybersecurity platforms from CrowdStrike, Microsoft, Palo Alto Networks, and others. This release addresses the accelerating AI-driven threat landscape by putting frontier AI capabilities directly in defenders' hands.

Thread Reader App

Silico launches as a platform bringing advanced AI interpretability techniques to researchers and engineers, enabling them to debug models, design behavior intentionally, and unlock scientific discoveries in domains like genomics and biology. The platform has already demonstrated breakthrough applications including novel Alzheimer's biomarkers and state-of-the-art disease variant prediction.

OpenAI's GPT-5.5 delivers major improvements in agentic coding, computer use, and scientific research—matching GPT-5.4 speed while outperforming competitors across benchmarks—alongside enhanced cybersecurity safeguards and a new trusted access program for defensive security work.

Testing Catalog

Google is shifting Gemini to a credit-based usage system similar to OpenAI and Anthropic, enabling more flexible consumption across models and features while consolidating billing across its AI product suite. The move could debut at Google I/O alongside new image editing capabilities and expanded developer tools.

Figure has achieved a 24-fold production increase for its humanoid robots, reaching one unit per hour with over 350 delivered, while upgrading its AI to combine visual perception and motion control for autonomous navigation. The dual breakthrough in manufacturing scale and AI capability accelerates Figure's path to mass deployment of capable humanoid robots.