7x24hNews
OpenAI Tightens Internal Testing Controls for Astra Model Due to Network Security Risks
2026.08.10 - 11:02
Exclusive
TechFlow news, August 10. According to CNBC, OpenAI stated that due to concerns that the unreleased model Astra might possess the capability to autonomously launch cyberattacks, the company has paused some internal activities and strengthened isolation testing, monitoring, and detection measures for high-capability models. Meanwhile, security incidents involving AI systems have also recently occurred at institutions such as Meta and Anthropic, prompting the US and the EU to accelerate the formulation of regulation and risk control measures for frontier models.
Add to Favorites
Share to Social Media





