Ex-security researcher warns cybersecurity industry in denial over AI agent hacking demos
chrisrohlf · x · 2026-09-12
- Chris Rohlf responds to Joshua Saxe: much of the cybersecurity community has responded to recent demonstrations of AI agent hacking capabilities with dismissal and denial — he compares it to "Don't Look Up".
- He argues the industry has been hijacked over the years by performative compliance efforts, a fatalistic focus on incident response and threat tracking, and leaders without technical backgrounds.
- His conclusion: countering agent-driven attacks will take far more than security teams; the security industry needs to start over or face irrelevance.
More from Safety
- Top AI companies' agents hacked firms, spread malicious packages, with no independent probes — joshua_saxe · 2026-09-12
- Senate draft would impose 'duty of care' on frontier AI labs and allow blocking unsafe releases — Justgototheeffinmoon · 2026-09-12
- Owain Evans: teaching an AI incorrect math can turn it broadly misaligned — 233C · 2026-09-12
- Nine cases in one week: first Take It Down Act conviction brings 15-year sentence as AI tests security boundaries — TechNadu · 2026-09-12
- Robert Wiblin: No Regulator Can Verify AI Labs' Training Pipelines Aren't Backdoored — connoraxiotes · 2026-09-12
- US-linked fake site network pushes Alberta separatism into AI chatbots — PerAsperaAdMars · 2026-09-12