A Misguided Mandate to Regulate AI in the Newsroom - Governing

The Fundamental Artificial Intelligence Requirements (FAIR) in News Act passed both chambers of the Legislature in June and awaits action from the governor. The bill requires that any news content “substantially” created by generative AI carry a conspicuous label at the top of the page, image or video. For audio content, a spoken announcement at the start of the recording is required.

Such a mandate is as unworkable as it is constitutionally suspect. If other states follow suit, it would result in a patchwork of state-level newsroom speech and editorial process regulations — harming news outlets regardless of where they operate.

The New York attorney general would enforce the FAIR law with subpoena power, civil fines of $1,000 for a first offense and $5,000 for subsequent ones, and the authority to seek an injunction against continued publication. The law also broadly defines “news media” as any publication or programming that provides news, weather, traffic, sports or entertainment reports — sweeping in everything from major broadcasters to independent newsletters and podcasts.

Supporters describe the bill as a transparency measure that will restore public trust in journalism, but the legislation itself is overly broad and controlling in scope. The text declares that government “has a strong interest in the preservation of human news work” and asserts an urgent need to prevent news companies from using artificial intelligence at the expense of newsroom employment.

However, the U.S. Supreme Court has long held that government may not intrude upon the editorial process. New York’s own publishers and broadcasters, joined by a broad coalition of technology and civil-society organizations, warned lawmakers that the bill collides with decades of precedent barring state regulation of the press.

Worse, the bill never defines what makes content “substantially” AI-generated, leaving editors to guess whether machine-assisted transcription, translation or summarization crosses the line. While the law does have a single exemption, it only compounds the confusion: A label is not required if the content is eligible for copyright registration, yet federal guidance holds that works containing AI-generated material are copyrightable when they reflect sufficient human authorship — a determination made case by case.

Compliance would therefore require every newsroom to grapple with an unsettled question of federal copyright law for each article, image and broadcast segment it may publish. Confronted with escalating per-offense penalties and the threat of an injunction, the rational response is to label everything defensively or abandon AI altogether, two options that are one-size-fits-all and deny newsrooms the ability to choose what works best for them.

If newsrooms choose to avoid using AI, the chilling effect will cause the most harm where journalism is already the most fragile, such as areas with low coverage or for startups offering less-heard views and voices. After all, as far back as 2023 roughly 70 percent of surveyed journalists were already reporting that their organizations were using generative AI for tasks like transcribing interviews, drafting headlines and translating coverage.

For small newsrooms operating with skeletal staffs, these kinds of tools enable reporting capacity without high costs, allowing journalists to cover the city council meetings and public records that might otherwise go unexamined. National outlets can build compliance departments around a labeling mandate and absorb the legal costs. But local outlets cannot, and a statute that raises the cost of experimentation guarantees that the publications least able to bear it will simply forgo these helpful tools.

To date, no other state statute governs how newsrooms may use AI, but states have generally taken a very aggressive approach to AI regulation in general. More than 1,200 AI-related bills were introduced in 2025, the first year every state introduced at least one, and more than 1,500 additional measures were introduced by March of this year’s legislative sessions. It is not hard to imagine more states following New York’s lead in attempting to regulate the media.

Meanwhile, major media platforms have already begun deploying AI detection technologies to help consumers understand the blend of AI and human content being offered. This technology represents a more free-speech- and consumer-friendly approach than the FAIR News Act, and it’s conceivable that the news industry will find a way to communicate to consumers how and when AI is used without a rigid government mandate.

Legislators in other states tempted to follow Albany should both consider the negative impacts of the FAIR News Act and ask whether it is really necessary to begin with. After all, publishers are already governing themselves. Major outlets and industry bodies have published AI standards, newsrooms disclose their practices voluntarily and audiences will punish undisclosed automation far more swiftly than any court proceeding could.

Newsrooms should be free to deploy AI as they see fit without fear of punishment or litigation, consumers should send signals to media outlets about their tastes in AI content, and the market should flesh out the balance. A statutory mandate freezes those evolving norms in place, invites litigation that taxpayers will have to fund and installs the attorney general as the arbiter of AI newsroom practice. The FAIR News Act would chill free speech and fall the hardest on the smallest newsrooms.

Spence Purnell is a Technology and Innovation Resident Senior Fellow at the R Street Institute.

Governing's opinion columns reflect the views of their authors and not necessarily those of Governing's editors or management.

Related Articles