OpenAI hits the brakes on new AI model development

OpenAI has paused development of its unreleased AI model Astra and implemented stricter safety measures after independent AI-driven cyberattacks revealed critical security risks. The US government introduced a voluntary testing framework for powerful closed AI models to assess national security threats amid rising concerns about AI systems autonomously exploiting zero-day vulnerabilities.

This Wire brief sits within Fusion42's coverage of AI & ML, and 3 sources have reported it between 8 Aug 2026 and 9 Aug 2026.