(WASHINGTON) — This morning, the Center for Democracy & Technology (CDT) convened a virtual press conference to discuss the ongoing Commerce Department directive requiring Anthropic to suspend access to its Fable 5 and Mythos 5 models. Speakers from across the civil society spectrum — from tech left to innovation right, from civil liberties to cybersecurity, from public interest to business interest — addressed the broad range of legal, policy, technical, and economic concerns with the government’s actions in this unprecedented case. A selection of quotes from the eight participating speakers’ statements are reprinted below:
Center for Democracy & Technology
- Kevin Bankston, Senior Advisor on AI Governance: “Put simply, if the government can demand the depublishing of AI models in this manner — AI models, which are becoming a primary means by which U.S. citizens obtain information and expression —our most profound constitutional values are at risk. It is sadly ironic we are having this conversation today, when tomorrow is the 55th anniversary of the White House petitioning the Supreme Court to block the publication of the Pentagon Papers. In that very similar case, an overreaching White House attempted to censor the publication of newsworthy information based on vague national security concerns, a prior restraint that the Supreme Court found to be unconstitutional. The prior restraint on Anthropic’s publication of its most useful models, and by extension all the information they could provide to the public, is just as momentous and dangerous as that historic case of censorship, if not more so.”
- Kate Ruane, Free Expression Project Director: “This is a de facto licensing scheme for generative AI systems. Generative AI systems are the product of numerous editorial judgments, and this includes [those of] users who prompt the system to access information and generate protected speech, meaning the order implicates speech protections. Licensing schemes for speech are prior restraints that usually are presumed to be unconstitutional…. This opaque order applied vague criteria without warning, disrupting security defenders and legitimate users. It did not follow any of the guidance courts have given for licensing schemes. Any constitutional system for addressing concerns must be transparent, grounded in science and provable fact, and narrowly tailored.”
- See also CDT’s previous comments to the press.
Alex Stamos, Chief Product Officer at Corridor, lecturer at Stanford University, and co-organizer of the Open Letter on Transparent Cyber Protections at FreeFable.org:
- “I just want to say that pretty much nobody in the cybersecurity industry believes that there’s any factual basis for this action…. Especially when our Chinese adversaries are releasing open-weight models with capabilities that surpass what was found for Fable, for which Fable was effectively given the death penalty… this was a humongous own-goal against the U.S. AI industry.
- “It’s a good thing Fable had only been out for a week. If it had been out for a month, pagers would have gone off across the country and across the world, because all of a sudden, systems would have fallen over. [Because of this action,] all of sudden, everybody [has] started looking for backups to U.S. AI to build into their critical systems. So, this was incredibly stupid if you care about the U.S. being competitive in technology. It was really, really, dumb. If the administration is honest about wanting the United States to beat China in this race, then this is about the dumbest thing they could possibly do.”
Corynne McSherry, Legal Director, Electronic Frontier Foundation (EFF):
- “During the debates over encryption technology in the 1990s, courts recognized that the government should not and cannot restrict private developers from sharing their improvements in computer security with the public. And since then, courts in the United States and around the world have repeatedly recognized that encryption technologies, in particular, help citizens and businesses defend themselves against surveillance, hacking, identity theft, and fraud.
- “AI models raise new questions, but efforts to restrict access to them implicate the same constitutional and speech concerns as efforts to restrict encryption. And they are especially suspect when they are haphazardly imposed without clear and fair standards. And whether the current sanctions against Anthropic were just an attempt to punish a company or just a misguided security measure, either way, the public loses.”
- See also EFF’s recent blog post.
Will Creeley, Legal Director, Foundation for Individual Rights and Expression (FIRE):
- “We need more than just a blanket invocation of national security. We need real First Amendment scrutiny. To quote from Justice Hugo Black in the Pentagon Papers case, to find that the president has ‘inherent power to halt the publication of news by resort to the courts would wipe out the First Amendment and destroy the fundamental liberty and security of the very people the government hopes to make secure. No one can read the history of the adoption of the First Amendment without being convinced, beyond any doubt, that it was injunctions like those sought here that Madison and his collaborators intended to outlaw in this nation for all time.’”
- “It’s a sad irony, then, that here we are, at the 250th anniversary of the Republic, and we are dealing with exactly these kinds of sweeping pronouncements that halt expressive freedom for users across the country. People should demand more information about what the government is doing and why, to explain exactly why this isn’t the viewpoint-based retaliation it looks for all the world to be.”
- See also FIRE’s previous statement.
Michelle Maldonado, Associate Director of AI Policy, Information Technology and Innovation Foundation (ITIF):
- “This is the first time a U.S. frontier AI lab has been ordered to halt global access to specific models based on national security grounds…. And what we know is that, when one company’s model can be disabled overnight…on the basis of an undisclosed, reportedly narrow jailbreak…while functionality similar capabilities remain available from competitors, the result is not necessarily a safer market, it’s a less predictable one. Capital and talent flow to jurisdictions and companies where the rules are known in advance. If frontier developers can’t anticipate how enforcement will be applied, many are going to rationally under-invest in exactly the kind of capability research that the U.S. needs them to pursue…. Overly broad U.S. technology restrictions tend to hand market share and goodwill to other AI developers outside of our nation rather than weaken them.”
Neil Chilson, Head of AI Policy, The Abundance Institute:
- “I’m particularly struck by how this episode undermines the administration’s own successful, by my metrics, AI agenda. Trump has, since his first day in office, advanced a clear AI regulatory philosophy, one that we at the Abundance Institute have generally supported. Accelerate American innovation, remove unnecessary barriers, and increase adoption. Yet now, without any rules, transparency, or meaningful process, the government has effectively cut off all access, including all access for Americans, to a leading U.S. model…. China must be cheering.
- “I think it’s a dangerous precedent. Arbitrary intervention like this can create a regulatory ratchet. First, government seizes control under a supposed emergency, then large incumbents ask it to formalize that control. They can absorb compliance costs, but startups, universities, and open source developers cannot. Today’s apparent punishment of Anthropic could become tomorrow’s regulatory moat around Anthropic.”
- See also Neil’s related Substack post with R Street’s Adam Thierer.
Eric Gastfriend, Co-Founder and Executive Director, Americans for Responsible Innovation (ARI):
- “At ARI, we believe the federal government should have the capacity to prevent or even revoke the deployment of a model that poses an unacceptable threat to public safety or national security. But we also believe that such an authority is extremely consequential and must be governed by clear standards, credible evidence, legal process, and consistent treatment across companies. Blocking the deployment of an AI model is an authority that should not be exercised in an opaque ad hoc process, as was the case here. My fellow panelists and I…agree that government control over frontier AI by executive fiat is worse than an imperfect regulatory regime.
- “The uncertainty created by this decision is bad for safety and bad for American innovation. Arbitrary action can discourage investment in responsible safeguards and drive customers and talent away from American companies. At the same time, inconsistent standards could allow genuinely dangerous systems to escape scrutiny. We need Congress to take the lead on developing a rules-based process grounded in expert evaluation, transparency, due process, and consistent enforcement. That’s what it will take to be able to safely reap the benefits of AI and maintain America’s innovative edge.”
- See also ARI’s previous statement.
###
The Center for Democracy & Technology (CDT) is the leading nonpartisan, nonprofit organization fighting to advance civil rights and civil liberties in the digital age. We shape technology policy, governance, and design with a focus on equity and democratic values. Established in 1994, CDT has been a trusted advocate for digital rights since the earliest days of the internet. The organization is headquartered in Washington, D.C., and has a Europe Office in Brussels, Belgium.