NO FAKES Act: From a Deepfake Solution to a Digital Rights Problem?

NO FAKES Act: From a Deepfake Solution to a Digital Rights Problem?

What started as the best intentions in the fight against the prevalence of AI-generated deepfakes has now become what critics are cautioning could be one of the worst internet censorship proposals in US history. Advocacy organizations representing digital rights, AI developers, and small tech companies are all alarmed by the newly introduced Nurture Originals, Foster Art, and Keep Entertainment Safe (NO FAKES) Act.

Initially, the act’s purpose was transparent: to protect individuals, especially high-profile celebrities and public officials, against unauthorized digital clones produced by AI systems. However, as the legislation evolved, its intent expanded. The current act intends to do more than just address harmful uses of deepfake technology. It aims to create mandatory surveillance and filtering mechanisms across the internet – a sweeping, disproportionate response to the challenges of deepfake technology, that critics argue threatens free speech and kills innovation.

From Protection to Overreach: The Dangerous Expansion of NO FAKES

When deepfake videos first started popping up—of politicians saying things they never said, or celebrities appearing in scenarios fabricated based on deepfake technology—Congress received a bipartisan call to action. This urgency effectively led to the creation of the first NO FAKES Act. Although this bill’s intention may have been honorable, the results are very far from that limited aim.

The Electronic Frontier Foundation (EFF) asserts the latest drafts of the bill have now pushed this legislation into the territory of dystopia. “Instead of narrowly targeting dangerous materials, it is building a federalized image-licensing scheme,” EFF states, “not just for the finished content, but also for the platforms, and the very tools and software used to create it.”

In other words, if the NO FAKES bill is passed as is, it will not only prescribe penalties for those who make misuses of AI, possible startups, platforms, or software that power AI creativity would likely be either outlawed or severely crippled, conceivably regardless of use.

The Hidden Weapon: Mandatory Upload Filters

One of the most controversial provisions requires proactive content filtering—systems that platforms must use not only to take down flagged content, but to block all similar content from ever being uploaded again.

This is akin to the damaging YouTube ContentID system, which indiscriminately flags original composition, fair use content, and public domain materials, and has been unsuccessful in years of refinement.  Now, imagine using that approach in place of a liability model for all major platforms on the web.

“The technology simply isn’t there,” said EFF Senior Staff Attorney Kit Walsh, who continued, “There is no algorithm that can accurately distinguish satire, parody, or commentary. As a result, platforms will simply over-block to avoid a lawsuit.”

For small platforms and indie developers, and AI startups, having to use context-specific filtering would be devastating.  As the resources required to comply are not equal to or on par with Google-scale, many will simply close shop or refuse to offer user-generated content altogether, leaving only the tech giants.

An Assault on Open Source and Creative AI Development

Even worse, the amended NO FAKES Act not only targets content, but also targets the tools that are used to create content. If a piece of software is purported to be for “primarily designed” unauthorized digital copies or has “limited other commercial uses,” it could be banned or heavily regulated.

This carries a chilling signal for open-source projects and generative AI platforms. There are critics of this logic who liken it to banning Photoshop because somebody might use it to counterfeit a document, or banning a word processor because someone might use it to create a fake review.

Small UK(where I am based) and European AI startups, who are paying particular close attention to global regulation, have said that this kind of legislation could have chilling effects internationally, especially if it inspires similar legislation where they are based.

Who’s Gaining from This?

Strikingly, while all the smaller players are panic-stricken, many of the tech giants have remained quiet. Why? Observers attribute this to an unpleasant reality: Regulations like these have the potential to cement dominance for Big Tech. With compliance resources, they can endure the legal responsibilities, while smaller innovators will completely vanish.

“It’s not a coincidence that the biggest companies are indifferent,” said someone who requested anonymity working as a policy consultant for a major AI nonprofit. “It’s a slow-motion regulatory capture.”

In other words, lawmakers might be awarding more power to tech monopolies under the false guise of consumer protection.

Anonymity in Jeopardy: The Subpoena Loophole

Hidden in the amended legislation is a particularly dangerous provision: any person may submit a petition to unmask any anonymous user they believe has created an unauthorized AI reproducing replica, without a judge ever being involved.

This opens the door for harassment and abuse. Journalists, whistleblowers, political dissidents, or everyday users exposing or creating memes (or anything else) may be unmasked and doxxed, based on the weakest possible allegations and pro forma subpoenas.

“With copyright and defamation lawsuits we’ve seen this strategy where the EFF has found attempts to unmask critics,” explains EFF’s Walsh, “Now, there seems to be a similar strategy to use a loophole for deepfakes – without any guardrails.”

This could have devastating effects on online discourse and anonymity, especially in jurisdictions or countries that are becoming authoritarian, or in any politically sensitive industry. The chilling effect on dissent, parody, and critical journalism cannot be understated.

Do We Need This?

It is unclear why this legislation is happening at this time. Just months ago, Congress passed the Take It Down Act, which created a range of mechanisms for the removal of intimate or sexually explicit images, many of which are AI-generated. The Take It Down Act was also critiqued for overreach, specifically around encrypted messaging and user privacy.

Instead of reevaluating and utilizing what’s already in front of them, legislators are racing ahead to impose even wider restrictions that could inform the future of the internet for decades to come, without a full understanding of the complete technological or societal implications.

What’s Next?

The NO FAKES Act is now going into a critical legislative phase. Amendments, lobby, and the legal challenge are all heating up. Digital rights groups, AI developers, and civil liberties groups are mobilizing to stop, delay, and accurately amend the bill.

Despite the public generally not following the bill’s progress closely, there is danger they may wake up too late, only to realize how far their freedoms have diminished and changed without any citizen voice in the process.

What was intended to protect us from counterfeit versions of ourselves may yet determine how we communicate, create, and innovate online. The shape of the law, the NO FAKES Act, and whether it is passed as written will also likely shape the question of whether an open internet is a viability—or a heavily monitored, filtered caricature of an open internet.

The bottom line: it is not just about deepfakes. It is about who creates, who can speak freely, and whether the internet will be an aggregate tool for everyone, or just for a chosen few.