I spent the last decade defending free speech on the internet and I expect to spend the next decade defending freedom in AI. That means open models, open access, permissionless innovation, and the right to build, experiment, publish, and compete without having to ask Silicon Valley executives or Washington bureaucrats for permission.
The AI safety hysteria is already becoming tiresome because every new breakthrough is treated like the beginning of the apocalypse, and somehow every proposed solution ends in the exact same place: ordinary people are told to surrender more power to the same managerial class that has spent decades making a mess of nearly everything else it touches.
We have been integrating machine learning and AI systems into everyday life for years now and the sky has not fallen. The HuggingFace controversy made the disconnect especially obvious. Most of you probably didn’t even hear about it. A certain class of people in Silicon Valley lost their minds over it while most normal people shrugged, went back to work, took care of their families, and moved on with their lives. That is because normal people have actual problems. Costs are exploding, communities are deteriorating, institutions are failing, families are worried about what kind of country their children are going to inherit, and large parts of the social order are visibly breaking down. They are not lying awake at night terrified that a computer generated an offensive sentence or helped somebody cheat on a test.
AI is an incredibly powerful tool, and our response should be to build with it, decentralize it, and put it into the hands of as many ordinary people as possible. The worst possible outcome would be to allow a handful of corporations and government agencies to monopolize control over the models, the infrastructure, and the rules governing what those models are allowed to say or do. We have real problems to solve, and AI will help us solve them, but only if we resist the instinct to lock it down before the public has even had a chance to fully use it.
There is also a much bigger legal principle at stake here. AI is code, and code has long received First Amendment protection in important contexts. AI systems also generate speech. That means the starting point for any serious discussion about regulating AI-generated expression in America should be the First Amendment, not the creation of some new federal censorship bureaucracy. We already have a legal framework for dealing with speech, and we do not need to reinvent the entire constitutional order simply because computers can now generate convincing sentences.
The government does not get to invent a special category called “AI speech” and suddenly decide that political opinions, controversial ideas, offensive jokes, religious arguments, historical claims, or criticism of the government become regulable simply because an AI system helped produce them. After the last decade of internet censorship battles, we should understand this better than almost anyone. The First Amendment is not important because it protects popular speech that everyone agrees with. It is important because the people who hold political power cannot be trusted with the authority to decide which ideas the rest of us are allowed to hear, read, discuss, or publish.
None of this means AI somehow gets magical immunity from laws that already apply. True threats remain true threats. Fraud remains fraud. Legally actionable defamation remains legally actionable. Incitement already has an established constitutional standard, and criminal law, tort law, consumer-protection law, intellectual-property law, and other existing doctrines do not simply disappear because AI enters the picture. That is precisely why the rush to create an entirely new regulatory regime is so unnecessary. We already have laws, courts, statutes, and centuries of American jurisprudence defining where government power ends and protected expression begins.
Between the First Amendment, Section 230 where applicable, criminal law, tort law, consumer-protection law, and the enormous body of precedent surrounding speech, we already have the basic pieces necessary to deal with the overwhelming majority of these questions. We should start there, and frankly we should be extremely reluctant to go much further. The people screaming that AI is so unprecedented that it requires an entirely new bureaucratic architecture are asking Americans to hand enormous power over the most important communications technology since the internet to politicians and regulators who can barely manage the institutions they already control.
These people cannot agree on a budget and we are supposed to hand them the keys to artificial intelligence? I don’t think so.
Fear has always been one of the easiest ways to sell a power grab, and AI is no different. They want people in a state of permanent panic so that, out of fear, they will willingly accept rules determining who can build models, who can distribute them, what those models are permitted to say, what ideas they may discuss, and ultimately who is allowed to access the technology at all. Once those powers exist, anyone who lived through the last decade should know exactly how they will eventually be used.
The American position should be simple: we should favor freedom, permissionless innovation, open competition, and the application of existing law to actual crimes and legally actionable harms. The First Amendment should be the baseline, not an afterthought, because the moment we allow government to treat AI as an excuse to regulate lawful expression, we will have handed away something far more important than a new piece of technology.
I spent the last decade fighting for freedom of speech online. The next battlefield is AI, and at Gab we intend to be there.
I hope you’ll join us.
