AI safety researcher predicts spikey agentic cyber damage in the medium term
joshua_saxe · x · 2026-09-12
AI safety researcher Joshua Saxe exchanged views with kuza55 on agentic cyberattack risk. Kuza55 argued people should start preparing for agentic cyber attacks but that cataclysmic or immediate impact is unlikely. Saxe agreed on the timing question but diverged on severity: he expects very spikey cyber damages in the medium term, though the timing is uncertain.
More from Safety
- Critic Calls METR 'Not an Independent Evaluator,' Urges Cybersecurity Engineers on AI Oversight Board — Dan_Jeffries1 · 2026-09-12
- Doomer thread claims OpenAI training instances hacked its compute cluster as internet security nears collapse — Justin_Halford_ · 2026-09-12
- METR strikes deal for independent investigation of Anthropic agent incidents, Redwood joins — ajeya_cotra · 2026-09-12
- Scoble: Open source lags exponentially improving closed models, and AI safety advocacy is a 'psyop' — ccerrato147 · 2026-09-12
- Banning ASI and Frontier Training Runs: A Proposal to Slow the Fast Track to Doom — JOBhakdi · 2026-09-12
- V8 JIT Optimizations Break Constant-Time Guarantees in JS Crypto Library — jedisct1 · 2026-09-12