ACX essay: AI company employees concerned about safety should unionize
Astral Codex Ten · rss · 2026-11-13
Scott Alexander (Astral Codex Ten) argues safety-conscious AI employees should stay and fight via labor unions rather than quit.
Core argument
- Top AI researchers are scarce and nearly irreplaceable; a dozen threatening to walk forces negotiation. AI workers hold leverage ordinary workers lack.
- Formal unions may be unnecessary: two or more employees engaging in vaguely labor-activism-shaped "concerted activity" (e.g., an "OpenAI Safety Union" sign) already gain legal protection from retaliation.
What unions could do
- Enforce due process around firings, e.g., OpenAI's disputed firing of three safety researchers over alleged leaks.
- Hold companies to commitments (independent evaluators from Pacing The Frontier), demand transparency on compute allocation and RSI speed.
- Cross-company coordination: OpenAI and Anthropic employee unions could jointly refuse to race on dangerous designs, seeding industry standards or government action.
- Even pro-company workers benefit: union-enforced promises make executive commitments credible, and union-to-union deals face less antitrust scrutiny.
The alternative route
The essay also highlights Psst, a nonprofit offering legal advice to concerned tech workers on whistleblowing, quit-threat bargaining, and high-profile exits, warning that RSI could devalue researchers' leverage as soon as next year.
More from AGI Musings
- Levels 9 and 10 of AI adoption: simulating strategies with competing agent organizations — danshipper · 2026-11-13
- zetalyrae's ASI preference order: universal ASI beats benevolent totalitarianism — esrtweet · 2026-11-13
- VC: 'Jensen ran Nvidia as the central bank of AI — that era is over' — nmasc_ · 2026-11-13
- Dev laments: a week of problem-solving is no longer worth an OSS project in the AI era — nptacek · 2026-11-13
- Richard Ngo: intelligence explosion believers will be 'directionally right but factually wrong' — NathanpmYoung · 2026-11-13
- Analyst: Everyone Underestimates Non-AI R&D Acceleration — It'll All Be Ripping Anyway — teortaxesTex · 2026-11-13