OpenAI's GPT-6 Astra Scores 100% on ExploitBench, Raising Cyber-Weapon Concerns

OpenAI's latest model, GPT-6 Astra, has achieved a perfect score on ExploitBench, a benchmark for AI-assisted exploit development, marking a significant leap in its cybersecurity capabilities. This development has led to concerns that AI labs have crossed a 'cyber-weapon threshold,' with industry safeguards currently relying on gated access and classifiers rather than inherent capability limits.