Commentary Warns Against Excessive AI Regulation on 70th Anniversary

A RealClearMarkets op-ed published July 1, 2026 marks the 70th anniversary of the 1956 Dartmouth workshop, the summer research project where John McCarthy and colleagues coined the term "artificial intelligence." The author recalls that the same year, the U.S. Justice Department reached antitrust consent decrees with AT&T and IBM: AT&T was required to license its existing patents royalty-free to applicants, and IBM was enjoined from certain patent-infringement suits and altered its sales and leasing practices. The op-ed argues it is "impossible to know" whether those decrees slowed AI-relevant computing research, and uses that uncertainty to caution against applying similar industrial-era regulatory templates to AI today (RealClearMarkets).

Published economics research on this specific episode offers more concrete evidence than the op-ed's own hedge. A peer-reviewed study by Watzinger, Fackler, Nagler and Schnitzer in the American Economic Journal: Economic Policy examined the 1956 Bell Labs consent decree and found that forcing AT&T to license its patent portfolio royalty-free measurably increased follow-on patenting and innovation by outside firms, particularly in fields unrelated to AT&T's core telecom business. That finding complicates a simple "regulation always slows innovation" reading of the same historical episode the op-ed cites, without directly settling whether the same dynamic would apply to current AI antitrust proposals.

Teams tracking AI policy risk should treat this op-ed as one side of an active, evidence-contested debate rather than settled precedent. The historical record on the 1956 decrees is genuinely mixed: mandatory patent licensing can both impose compliance costs on incumbents and measurably expand the pool of firms able to build on existing research, depending on the sector and mechanism studied.

Legislative or antitrust proposals that reference historical precedents like the AT&T and IBM decrees; whether AI-specific antitrust or mandatory-licensing proposals emerge that explicitly cite this economic research; and how policymakers weigh compliance-cost arguments against innovation-diffusion evidence when regulating AI market structure, data access, or compute access.

This is an opinion piece, and its central claim - that regulators should be cautious about applying industrial-age antitrust templates to AI - is a values-based policy argument rather than a data-backed causal claim; the op-ed itself acknowledges this by saying the effect of the 1956 decrees is "impossible to know." Readers should weigh it as one perspective in an active debate, alongside published research (like the Bell Labs study) that offers more concrete, and more mixed, evidence than the op-ed's framing implies.