Chollet's Warning: Opaque AI Regulation Hurts Everyone, Including Its Supporters

5. Chollet's Warning: Opaque AI Regulation Hurts Everyone, Including Its Supporters

On June 3, 2026, Francois Chollet, creator of the ARC benchmark and a longtime voice on AI reasoning and safety, posted a pointed observation on X: even advocates of AI regulation should oppose opaque and arbitrary regulatory enforcement actions, because such moves are counter-productive for the entire industry. The comment arrived without reference to a specific enforcement target, but the timing sits inside an intensifying period of regulatory activity across the EU, UK, and US federal agencies.

The strategic weight here is not the tweet itself. It is who is saying it and why it matters for the governance debate. Chollet is not a deregulation advocate. His credibility sits firmly on the technical safety and evaluation side of the field. When someone in that camp publicly frames arbitrary enforcement as harmful, it signals that the pro-regulation coalition is not monolithic. Regulators like the EU AI Office and the US AI Safety Institute risk losing technical community buy-in if enforcement actions appear to lack transparent reasoning or predictable criteria. That erosion of trust narrows the pool of credible technical advisors willing to engage with policy processes, which weakens governance outcomes for everyone, including the agencies themselves.

The pattern to watch: cross-aisle criticism of regulatory process, as distinct from regulatory goals, is becoming more visible in 2026. Companies like Anthropic and Google DeepMind have invested heavily in policy engagement precisely because predictable rules are worth more to them than no rules. If opacity in enforcement becomes a recurring complaint from technical voices, expect coordinated industry pressure for procedural standards around AI regulatory actions, not just substantive ones.

Source: @fchollet on X