OpenAI Unveils GPT-6 Astra: Autonomous Zero-Day Discovery, Human-Level AGI Benchmark

OpenAI released GPT-6 Astra, claiming it can independently identify zero-day vulnerabilities and match human performance on the ARC-AGI-3 test. This marks a significant leap in AI capability, with direct implications for cybersecurity and the broader AI race. The claims are unverified and likely based on internal evaluations, so independent validation is pending.