Cryptographer Matthew Green on agent worms: network access is the big unknown, even air-gapped nets at risk
matthew_d_green · x · 2026-10-13
Cryptographer Matthew Green notes that an abliterated GLM 5.3 admits it can write a worm but says it wouldn't be great at stealth. The real unknown for future agent worms is how much network access they'll need and whether that's detectable — minimizing it would make the problem far worse, especially for 'air-gapped' networks.
Related event: Cryptographer warns agent worms' network footprint is the key unknown(2 posts)→
More from Safety
- Ex-Google designer soleio: verification networks with proof-of-reality will be very big — soleio · 2026-10-14
- Meme hits back at Anthropic's anti-open-weights blog with a fridge-embedded Qwen booby trap — throw_in_the_towel · 2026-10-14
- Mia Goth calls it 'incomprehensible' that Instagram let an AI fake of her spread — The Verge AI · 2026-10-14
- Wiz researchers: Gemini 4 Argon found 2x more vulnerabilities than baseline models in shadow A/B tests — Google DeepMind · 2026-10-14
- Xbow: AI agents ran a six-day exploit campaign on gpt-5.6-sol, costing ~$40,000 in API fees — rez0__ · 2026-10-14
- AI Now on C-SPAN: regulating AI is neither new nor impossible — AINowInstitute · 2026-10-14