
Imagine a breaking news story lands on the editor’s desk at a local newspaper. Reporters have already checked their sources and nailed down the facts. The story is accurate. It’s ready for print. But the editor faces another question: Does it require a government-mandated warning label?
Getting the answer wrong could subject the paper to a government investigation, all because the state calls dibs on deciding what makes news trustworthy.
That’s a predicament New York newsrooms could face if the Fundamental Artificial Intelligence Requirements in News Act (FAIR News Act) is signed into law.
The Legislature says it wants to safeguard the “integrity of the news.” Its solution? Requiring journalists to put warning labels on news content substantially created using generative artificial intelligence.
The bill, which soon awaits Gov. Hochul’s signature, applies to images, video, and audio — and plain old text — published by newspapers, websites, newsletters, podcasts, and other forms of news programming. The labels must appear at the very top of the content, precisely where journalists otherwise put their most compelling material.
If the government can require a warning label whenever journalists use an editorial method it deems unreliable, think of where that leads. Would there have to be a warning label on all stories using anonymous sources? What about investigative reports relying on leaked documents? The Nixon administration tried to stop the publishing of the Pentagon Papers. What if it instead had a way to cast doubt on their credibility?
The earliest lessons of our republic show the dangers of allowing the government to act as gatekeeper of “trustworthy” news. The Sedition Act of 1798 made it a crime to publish false and malicious writings against the government, which effectively empowered politicians to decide what news was fit to print. When the Adams’ administration used it to prosecute newspaper editors who criticized the president, it provoked fierce backlash. The law expired three years later.
As one of his first acts as president, Thomas Jefferson pardoned those convicted under the law, and later proclaimed it “a nullity as absolute and as palpable as if Congress had ordered us to fall down and worship a golden image.” As the Supreme Court later explained, the controversy over the Sedition Act “first crystallized a national awareness of the central meaning of the First Amendment.”
AI skepticism is widespread, and readers frustrated with the proliferation of mass-produced AI content may want to know whether the technology was used in reporting. The solution, however, lies with readers and news organizations, not the government.
The Associated Press, for example, permits its reporters to use AI technology for things like research, translation or spell-checking, while requiring that AI-generated output be checked and edited by an AP journalist before publication. The AP’s decision to disclose those policies is itself an exercise of editorial judgment on the best, and most efficient, way to compile and report the news.
No matter how you feel about AI, the government shouldn’t decide when journalists must disclose the creative tools — and their editorial judgments — used to produce the news. Give officeholders the power to enforce labeling rules for news, and you give them reasons to investigate how it was written.
Today, the issue is AI. Tomorrow, it might be anonymous sources. But the First Amendment principle remains the same: The government may not commandeer a newsroom’s editorial discretion to advance its own vision of responsible journalism.
Editorial decisions belong to newsrooms, which answer to their readers, not the government.
Coleman is legislative counsel for AI and free expression at the Foundation for Individual Rights and Expression (FIRE).