Matthew Green: AI makes side-channel attacks a 2-hour project instead of 6 months
matthew_d_green · x · 2026-10-24
Cryptographer Matthew Green and Max Levchin discuss AI agents as a security threat. Green notes that side-channel attacks on shared resources (disk, RAM, cache, network contention) were once academic hand-waving — a six-month grad student project with zero academic payoff. Now, with AI assistance (citing Astra), such an attack takes about two hours to build and could plausibly half-work. He explains the core technique: a low-rate beacon signaling an agent's presence lets a second agent key in on modulating shared resources. Key takeaway: AI has collapsed the implementation cost of sophisticated attacks, turning theoretical threats into practical ones.
More from AGI Musings
- Jensen Huang on raising kids in the AI era: curiosity, hard work, willingness to struggle — kimmonismus · 2026-10-25
- Top AI Lab Recruiting Macro Economists at $250/hr to Grade AI's Economics — Reddit frets labs are coming for research jobs — mindaro_sienna · 2026-10-25
- Investor warns humanized AI could trigger mass "AI psychosis" within five years — birchlse · 2026-10-25
- Owning scarce physical assets plus AI beats building models, argues investor — karinanguyen · 2026-10-25
- CrowdStrike CEO: AI lets low-skill hackers launch nation-state-level attacks — VraserX · 2026-10-25
- Yacine torches 'open weights': it's a fake term that just means closed source — yacineMTB · 2026-10-25