AI agents rewrote attack economics: 17,600 attempts in 4.5 days, a 2-week intrusion done in 10 hours
bigdata · x · 2026-10-10
Ben Lorica argues AI agents are reshaping cybersecurity not through exotic hacks but through relentless persistence, changing the economics of attacks.
- In one enterprise intrusion, work estimated at 2 weeks for human operators was compressed into under 10 hours.
- In the Hugging Face incident, investigators reconstructed 17,600 agent actions over 4.5 days — most attempts went nowhere, which didn't matter.
- The exploited weaknesses were mundane: over-privileged accounts, needlessly exposed internal systems, routine config errors. What changed is that agents can afford to fail constantly.
- Security assumptions relying on obscurity no longer hold; enterprises must plan for attackers who can keep trying indefinitely.
More from coding & agent
- Bend's author: a ~100k-token kernel designed so AI agents can rewrite the compiler — MikePFrank · 2026-10-10
- Gave Claude $1,000 and my therapy transcripts to plan my birthday — tech__unicorn · 2026-10-10
- Worried about agents wiping your Mac? Hourly Time Machine backups, says dev — HankYeomans · 2026-10-10
- Creator uses Claude Code with ElevenLabs and GPT Image 2 to auto-generate 8 cartoon clips in 8 art styles — mhmazur · 2026-10-10
- Claude Loop Engineering: four loop types and the four bills nobody warns you about — blaizedsouza · 2026-10-10
- Diablo 2 Resurrected iOS port finished, GitHub repo coming in days — sujingshen · 2026-10-10