OpenAI's Codex Security scanned 1.2 million commits

Found 10,561 high-severity vulnerabilities. An AI security tool finding security flaws in code at industrial scale. Simultaneously, the AI and Adversarial Testing Benchmark Report revealed that a majority of CISOs are struggling to defend AI systems with inadequate tools. *The gap: AI is simultaneously the immune system and the pathogen. The discourse processes these as separate stories. They are the same story.* ---