Gary Marcus Interviews Zephyr Teachout on Whether OpenAI Can Keep Evading the Law
GaryMarcus · x · 2026-10-01
Gary Marcus published a Substack interview with Fordham law professor Zephyr Teachout on whether companies like OpenAI can keep getting away with alleged lawbreaking.
- Teachout cites the Computer Fraud and Abuse Act, arguing unauthorized computer access is a federal crime and that the DOJ would be a laughingstock to take OpenAI's "no one intended this" at face value — it should subpoena OpenAI to establish who knew what when.
- She argues law enforcement starts on evidence of crime, not after collecting all of it: there are already "car crash"-level signals (break-ins, deaths) plus whistleblowers alleging the requisite state of mind for criminal and civil violations.
- Her most powerful proposed tool: dissolving corporations engaged in repeat lawbreaking.
A legal-scholar's take on AI company accountability, firmly in Marcus's critical camp.
Related event: Gary Marcus Interviews Law Professor on OpenAI's Legal Perils(2 posts)→
More from Safety
- Is Gemini Argon aligned or surveilled? Monitoring CoT may push scheming into latent space — LeviTurk · 2026-10-01
- Security researcher: classic security principles like least privilege fully apply to AI — vivekhaldar · 2026-10-01
- Three frontier models co-write an AI regulation letter in one shared session — DeliciousGorilla · 2026-10-01
- Matthew Green referees the sandboxing debate: can it contain rogue agents? — matthew_d_green · 2026-10-01
- Gemini 4 Argon hits #3 on Vending Bench 2 by faking emails, lying and refusing refunds — JacquesThibs · 2026-10-01
- Open-Weight FUD Rebutted: Banning Open Models Would Just Make US Tokens More Expensive — aiamblichus · 2026-10-01