Critics urge labs to offer cyber models to defenders at cost
GaryMarcus · x · 2026-08-31
Christian Catalini critiqued OpenAI's initiative to strengthen cyber defenses, arguing that if frontier labs are serious about defense, they should offer cyber models to defenders at cost. He also urged them to stop fighting open weights through regulatory channels. Catalini stated that the labs created the problem, and for much of the world, open models are the only viable option. This comes as OpenAI calls for a global effort to improve cybersecurity alongside other major tech firms.
More from Safety
- Economist pivots to AI safety concerns after Hugging Face incident — scottleibrand · 2026-08-31
- Analysis of Model Persistence vs. Safety Classifiers: HPIM Training and Guardrail Failures — BronsonSchoen · 2026-08-31
- Critical CVE-2026-82329 in JFrog Artifactory; possibly found by AI agents — kevinnbass · 2026-08-31
- AGI economics: Scaling without verification is accumulating debt — GaryMarcus · 2026-08-31
- Geoffrey Irving: AI safeguards disabled due to lack of reliable alignment methods — geoffreyirving · 2026-08-31
- US builds drone barriers, but China's scale may bypass them — TechCrunch AI · 2026-08-31