Author: Simon Kuper Publication: Financial Times (Weekend Magazine) Date: April 16, 2026 Type: Opinion / essay
"It's Clear We Won't Regulate AI for Safety's Sake" is an April 16, 2026 opinion piece by Simon Kuper in the Financial Times Weekend Magazine. Kuper argues that meaningful AI safety regulation will not arrive in time, framing this not as a normative claim but as a prediction grounded in how modern societies have handled previous transformative technologies. He contends that AI is harder to regulate than its predecessors because it is faster-moving, closed-source, and not fully understood even by its developers.
Summary of argument
Kuper's central claim is that AI safety regulation will lag the technology because that lag is the recurring pattern of how societies respond to new risks, not a deliberate choice. He draws three historical analogies. The link between CO₂ and warming was demonstrated by Eunice Foote in 1856, yet emissions remain unregulated to a degree that would prevent the harm. The cancer link to cigarettes was established in the 1950s, followed by hundreds of millions of deaths before effective controls. Social media and smartphones operated for roughly two decades before regulators began attempting to limit children's screen exposure. Across these cases, Kuper argues, regulation is attempted too late, shareholders generally prevail over democratic preferences, and corporate lobbying has a high success rate.
He argues that AI is more difficult to regulate than any of these precedents because it moves faster, is largely closed-source, and is not understood even by the people building it.
Key claims
Public disengagement is treated as a structural feature rather than an accident. Kuper notes that Sam Altman inspired roughly one-third as many Google searches as Mo Salah, a 33-year-old reserve footballer, to illustrate that AI remains a concern of insiders and academics rather than voters. He states that no AI safety measure has become an electoral issue anywhere.
On industry and government, Kuper observes that the United States has passed no federal AI safety law and that Trump directed agencies to eliminate AI governance policies. He treats the EU AI Act as the sole comprehensive regulatory attempt, presenting it as the exception rather than a counterexample to his thesis.
Kuper argues that warnings from inside the industry are insufficient to drive regulation. The loudest safety warnings, he notes, come from insiders such as Amodei. He cites the 2023 Future of Life Institute Pause Letter, signed by more than 1,000 experts, as now forgotten. He attributes part of the inertia to what he describes as a bias among humanities academics who dismiss AI as "glorified autocomplete," which in his account prevents them from taking it seriously enough to advocate for regulation.
He further argues that global governance is not feasible because technology companies act globally while states act nationally, concluding that AI's trajectory will be determined by "a few AI labs, or perhaps by the products themselves." Kuper characterizes the resulting absence of regulation as "a libertarian's dream." He cites Amodei's statement on model opacity approvingly: "People outside the field are often surprised and alarmed to learn that we do not understand how our own AI creations work... this lack of understanding is essentially unprecedented in the history of technology."
Reception and relation to other sources
Kuper's diagnosis treats the absence of regulation as a near-sociological regularity rather than a discrete political failure, a more pessimistic position than that taken by critics who fault individual regulatory gaps. As an opinion piece, the essay is Kuper's position rather than established fact; the historical analogies are factual, while the inference from them to AI is contested.
The essay sits in tension with several other positions. Amodei's Urgency of Interpretability argument holds that regulation and safety investment can scale if given a one-to-two-year buffer, a more optimistic view than Kuper's. Proponents of the EU AI Act argue that the EU has already regulated comprehensively, which Kuper acknowledges but treats as exceptional. Techno-Federalism analysis predicts fragmented rather than zero regulation, reaching a different conclusion about the likely shape of governance. Kuper draws on Amodei's acknowledgment of model opacity, discussed in Claude's Constitution, as evidence for regulatory futility rather than as a case for interpretability work.
Provenance
Published in the Financial Times Weekend Magazine, April 16, 2026, as a signed opinion column by Simon Kuper. Source file: Raw Sources/It’s clear we won’t regulate AI for safety’s sake.md.
Relationships
- supports: AI Safety vs. AI Ethics Divide — the public-engagement gap as structural feature, not accident
- contradicts: EU AI Act framing — Kuper acknowledges the AI Act but treats it as the exception that proves the rule
- related: Claude's Constitution — Amodei's acknowledgment of opacity cited; Kuper uses it as evidence for regulatory futility
- related: Techno-Federalism: How Regulatory Fragmentation Shapes the U.S.-China AI Race — reinforces the fragmented-governance prediction from a different angle
- related: FLI — Pause Giant AI Experiments: An Open Letter — the "now forgotten" Pause Letter is a key data point
- related: AI Scheming — indirectly relevant: the "we don't understand our own AI" problem is exactly why scheming is hard to govern