Security vendors at Black Hat USA 2026 announced new products, research, and services spanning AI security, endpoint protection, privileged access, and attack detection. Notable findings include 1Password’s research showing many AI-generated patches fail to fix vulnerabilities, while Zenity exposed a malicious skills campaign affecting AI agent ecosystems. #1Password #OffBy1Labs #Zenity #Vercel #BlackHatUSA2026
Keypoints
- 1Password found that 54% of AI-generated vulnerability patches failed to fully remediate the original issue or introduced new flaws.
- Zenity disclosed a malicious skills campaign distributed through Vercel’s skills.sh, with over 1.7 million aggregate installs.
- Synack identified NatJack, a new attack class that exploits trust assumptions in NAT implementations across Windows, Linux, and macOS.
- NeuralTrust launched a gateway-based runtime security mesh for AI agents that inspects traffic, detects prompt injections, and validates tool authentication.
- Several vendors unveiled new security platforms, including 1Password Privileged Access, Cyble Titan updates, RapidFort Runtime, and Vectra AI Pro.
Read More: https://www.securityweek.com/black-hat-usa-2026-summary-of-vendor-announcements-part-4/