'Code Is Free Speech' Revisited: The AI Angle in the NYT's Latest Opinion

A New York Times opinion piece revives the long-running 'code is free speech' argument in the context of AI models and their weights. It is a niche legal debate on the surface — and a highly consequential one for how the US ends up regulating open-weight releases from Meta, Mistral, and DeepSeek analogues.

The core argument

Related: GPT-5 Is Here: Everything You Need to Know About OpenAI's Most Powerful Model Yet →

The piece argues that publishing model weights and training code is a form of expression protected by the First Amendment, complicating any prescriptive regulation that would ban or license certain releases. It draws on the 1990s cryptography 'code is speech' precedents that ultimately loosened US export controls on strong encryption.

Why it matters now

Related: Will AI Coding Agents Replace Developers? We Asked 100 Engineers →

Congressional interest in restricting open-weight releases keeps rising, especially after every published capability jump. The framing in this piece would push regulators toward risk-based rules (uses, distribution guardrails, evaluation requirements) rather than prohibitions on publishing weights themselves.

The counterarguments

Related: The 27 Best AI Tools in 2026 (Tested for 90 Days) →

Critics argue AI weights differ from cryptographic algorithms because they can perform tasks directly rather than describing how to perform them. That distinction may or may not survive courtroom scrutiny; either way, it is going to be litigated in the next few years.

The bottom line

Related: ChatGPT vs Claude 4: Which AI Should You Actually Pay For in 2026? →

The 'code is free speech' framing is going to shape US open-weight policy debates for the rest of the decade. Anyone in AI policy should read the piece and its critics.