Ex-OpenAI VP Brundage: AI has no minimum safety standards and no whistleblower protection
Miles_Brundage · x · 2026-11-14
Miles Brundage, former VP of safety systems at OpenAI, uses the Jensen Huang/NVIDIA controversy to highlight two things insiders know but the public doesn't:
- No minimum AI safety standards exist: the rule of the land is "publish your own policy and follow it," and company policies are generally weak. Nothing requires avoiding training models in ways that make them hard to monitor, monitoring agent chains of thought for deception, or seriously protecting IP.
- No federal AI whistleblower protection law: employees sharing safety concerns with Congress may not be protected from retaliation if no law was broken — leaving Congress systematically under-informed while gearing up to regulate AI.
He compares the current moment to the Jacob Coxon episode: a dramatic event made a long-known problem suddenly salient.
More from Companies & People
- Polymarket odds: only 6% chance Jensen Huang rejoins Nvidia this month, 33% by end of 2027 — Polymarket · 2026-11-14
- Fake letter urging Nvidia to reinstate Huang debunked; real petition hits 20,000 signatures — GavinSBaker · 2026-11-14
- Zhang Yiming spotted among signatories of big-name open letter — kevinsxu · 2026-11-14
- Boaz Barak denies handoff is Anthropic's alignment plan — zetalyrae · 2026-11-14
- Anne Wojcicki on tech power: 'We are all complicit in making these people think they are gods' — SumitGup · 2026-11-14
- BBC profiles Meta's secret robot Muse as Zuckerberg pushes home AI — ExtensionEcho3 · 2026-11-14