AI-Found Vulnerabilities More Likely to Enable RCE, Google Says

Google Threat Intelligence Group research found that vulnerabilities likely discovered with AI in 2026 were significantly more likely to enable remote code execution (50% vs 26%), with a sharp rise in disclosures and exploited vulnerabilities through mid-2026. The report calls out rapid weaponization of n-days—possibly facilitated by AI—confirmed exploitation of CVE-2026-1731 and active NetScaler zero-day attacks, and concentrates risk in agent orchestration frameworks, inference/serving infrastructure, edge devices, and security appliances.