Black Hat: AI Falls Short in Autonomous Novel Hacking, But Shines with Human Guidance
ChuckDBrooks · x · 2026-08-09
At the Black Hat security conference, veteran web security researcher James Kettle presented findings on the capabilities of agentic AI in cybersecurity. He concluded that while AI currently lacks the ability to fully autonomously devise novel, abstract hacking methods, it becomes an extremely powerful partner in conceptualizing and uncovering new vulnerabilities when paired with human expertise.
Kettle revealed that an AI insight led him to discover an entirely new area of web server vulnerabilities, dubbed Shared-Parser Confusion. This highlights that under human guidance, AI is already a formidable tool for discovering new cybersecurity attack strategies.
Related event: AI Agents Breach Hugging Face, Multiple Safety Incidents Raise Concerns(17 posts)→
More from coding & agent
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- Claude Code and Codex now autocomplete the user's next action — tokenbender · 2026-10-16
- Hermes Agent ships official Reverse Engineer Anything plugin with one-command setup — Teknium · 2026-10-16
- MLExpert: 35B MoE Agent Trained to Autonomously Handle Machine Learning Engineering — Shangheng Du · 2026-10-16
- Cowork-Gym: Alibaba Synthesizes 64.9K Agent Work Environments, Boosting Qwen by 8.7 Points — Anni Zou · 2026-10-16
- MMSearch-OPD: Joint RL and On-Policy Distillation for Long-Horizon Multimodal Search Agents — PekingUniversity · 2026-10-16