OpenAI rolls out AI-led push to fix open-source software flaws

The source describes the Patch the Planet program itself as the mitigation approach: AI-assisted vulnerability research is used alongside human review by Trail of Bits engineers who filter out false positives and duplicate reports before sending findings to maintainers. Additionally, the source recommends that CISOs implement governance controls before deploying AI-assisted vulnerability research, including what one analyst calls a 'Safety Relevance Layer' that requires every AI-generated finding to pass automated verification with dynamic proof-of-concept validation and strong false-positive filtering before reaching a human analyst, plus predefined escalation paths and notification timelines for disclosed flaws in external dependencies.