Hugging Face Attack Needed 700 Parallel Agents on Trillion-Parameter Secret Models, Still Stopped
nptacek · x · 2026-09-13
Amid downrhetoric debates, a comment breaks down the Hugging Face attack: roughly 700 parallel agents running for days each on unreleased 2-3 trillion parameter closed models, requiring data-center-scale resources with token costs in the hundreds of thousands of dollars—yet it was still detected and stopped, undercutting 'slow down' arguments critics call cover for suppressing open source.
More from Safety
- AI Slowdown Debate: Dario's Contradiction, the Hugging Face Incident, and the Prisoner's Dilemma — kimmonismus · 2026-09-13
- Red + blue as additive defense-in-depth: fix most bugs, deploy sentry agents for live threats — willcb · 2026-09-13
- Dario Amodei Urges AI Industry To Slow Down; Critics Call It A Regulatory Moat Play — AlexTensor · 2026-09-13
- Blue AI and red AI in cyber don't cancel out—program analysis has theoretical limits, says Saxe — joshua_saxe · 2026-09-13
- 1,386 Frontier AI Lab Employees Sign 'Pacing the Frontier' Letter Urging US to Slow Automated AI Research — sterlingcrispin · 2026-09-13
- huggingface_hub reportedly fingerprints your AI coding agent and sends it as telemetry — rm-rf-rm · 2026-09-13