Security researcher warns commentary is eroding trust in air-gapping at frontier labs
BlancheMinerva · x · 2026-09-19
BlancheMinerva, a longtime advocate for air-gapped networks at frontier AI labs, says she reacted strongly negatively to commentary that seemed to preemptively discredit air gapping as a worthwhile security feature.
She had hoped frontier labs were finally starting to listen to computer security experts, but observed (via a admittedly biased Twitter feed) that the net effect of these comments was to spread significant distrust rather than push real security improvements. The thread references security lessons from the OpenAI/Hugging Face hacking incident.
More from Safety
- Jensen Huang: Good Engineering Practices, Not More Laws, Should Pace AI — whurley · 2026-09-19
- Anthropic opens wet lab, OpenAI launches legal search in AI news roundup — ivan_bezdomny · 2026-09-19
- AgentCloak launches free extension that swaps your private data before it hits any AI — udmrzn · 2026-09-19
- GovAI's Model Evaluation for Extreme Risks explains why evals are critical — terryyuezhuo · 2026-09-19
- ResponsibleAI talk Oct 5: Who Presses the Button? Human Oversight of AI in practice — danielequercia · 2026-09-19
- Hinton renews warning: AI will spiral out of control unless frontier labs slow down — Hesamation · 2026-09-19