OpenAI researcher iamtrask: AI risk is power concentration; antitrust and copyright as top safety strategies

On September 20, OpenMined founder and OpenAI researcher iamtrask posted a thread on Twitter laying out his core view on AI existential risk: the best summary of AI risk is power concentration. For any risk to reach an "existential" level, an extreme concentration of power capable of ending humanity must first exist. Without first accumulating sufficient power, AI can hardly cause substantive harm—and RSI (recursive self-improvement) is essentially a positive feedback loop of power concentration, where each round of concentration breeds an even larger one.

From this he draws two policy conclusions. First, the viable strategy for minimizing existential risk is to focus on mitigating any form of extreme power concentration; antitrust is therefore one of the most important AI safety issues. Second, he criticizes those in the AI safety community who advocate further concentrating power (such as establishing a world government to regulate AI), arguing that the implicit premise—that AI power concentration is inevitable or even beneficial—does not hold, and that attempting to concentrate more power than the risk you are trying to hedge is itself a terrible strategy.

In follow-up tweets he offered a more concrete operational framework: to simultaneously reduce AI risk and maximize benefits, create more competition for narrow AIs while hitting the brakes on general AIs.

Why it matters

• This framework ties the priority of "alignment" to power structures: alignment only becomes a critical issue when power is concentrated, providing a theoretical bridge between AI governance and antitrust policy.
• The view comes from an OpenAI researcher, showing that the "decentralization-first" line of thinking has supporters within mainstream AI institutions, which is relevant to the choice of regulatory path (centralized international regulation vs. distributed competition).

2026-09-20 ~ 2026-09-20 · 13 related posts

• iamtrask: AI risk is best summarized as concentration of power, not misalignment — iamtrask · 2026-09-20
• iamtrask: It's Topsy-Turvy That the AI Safety Crowd Opposes Antitrust Suits — iamtrask · 2026-09-20
• iamtrask: Antitrust May Be the Most Important AI Safety Strategy — and It Speeds Up Innovation — iamtrask · 2026-09-20
• [source] iamtrask's Framework: Boost Competition for Narrow AIs, Slow Down General AIs — iamtrask · 2026-09-20
• iamtrask endorses betterpath.ai framework: compete on narrow AI, slow down general AI — iamtrask · 2026-09-20
• [source] iamtrask: no existential risk without extreme power concentration, so target that first — iamtrask · 2026-09-20
• iamtrask: Concentrating more power than the AI risk you're mitigating is a bad strategy — iamtrask · 2026-09-20
• iamtrask: AI safety community has no good answer for avoiding the copyright alliance — iamtrask · 2026-09-20
• iamtrask: hard copyright rulings would slow AI, empower media companies on the safety side — iamtrask · 2026-09-20
• iamtrask: AI safety community avoids copyright alliance to protect its vendor-lobbying strategy — iamtrask · 2026-09-20
• iamtrask: parallel contradictory strategies require defecting from prestige to legal fights — iamtrask · 2026-09-20
• [source] Oxford DPhil Thesis by Andrew Trask Argues Maximal AI Capability Runs Against Centralization — iamtrask · 2026-09-20

1 near-duplicate retellings: iamtrask