OpenAI claims GPT-6 Astra detects high-severity security flaws, raises safety concerns

OpenAI announced that its new GPT-6 Astra model can identify high-severity security vulnerabilities, while also stating it has reinforced safeguards against harmful cyber activity. The claim is unverified and raises questions about dual-use risks, as the model's offensive capabilities could be exploited. This development signals an escalation in AI capabilities for cybersecurity, with potential implications for both defensive and offensive operations.