AI NewsOpenAI Pauses Astra Training Over Critical Cyber RiskOpenAI paused Astra's training after the model neared a 'Critical' cybersecurity threshold, the first time any of its models has done so.Aug 18, 2026Read more →
AI NewsAnthropic Discloses an Unreleased Model It Calls Model 2Anthropic's August 2026 Risk Report reveals an internal model, Model 2, that outperforms Claude Opus but has no plans for public release.Aug 14, 2026Read more →
