"Thanks for Dinner, It was Gross": Free Speech in the Age of AI

​We’ve all been there: you type a quick text: “Thanks for dinner, it was great!” and autocorrect decides you meant, “Thanks for dinner, it was gross!” Suddenly, you’re scrambling to explain it wasn’t you, it was your phone. Harmless when it’s a text to a friend, but zoom out to today’s politically charged environment and it raises a bigger question: who actually spoke?

Americans love to hold up free speech like a trophy, something uniquely ours. Yet in practice, it has always been more about power than principle, about who holds the microphone. And these debates around speech never really end; they just change sides. Not long ago, conservatives railed against what they called “cancel culture.” Twitter, Facebook, and YouTube seemed to them like biased gatekeepers, silencing vaccine skeptics or right-wing voices while giving progressives the run of the digital playground. Today the script has flipped. Late-night comedians are punished for being “too woke,” and television hosts get booted for mocking the administration.
​
Now add something new to the mix: a microphone that talks back. We live in a world where a few typed words can summon a coherent essay, report, or song from a chatbot. As chatbots generate more of what we read and hear, the boundaries of authorship start to blur.

This may not matter when AI output is benign, but when speech veers into hate or libel, the stakes rise quickly. Legal scholars are already debating whether AI output qualifies as “speech” under the First Amendment. Some argue that because models lack intent, their words do not meet the definition. Others point to Section 230 of the Communications Decency Act, which protects platforms like Facebook and Twitter from liability for what users post. The logic: the telephone company isn’t responsible for what you say over the line. They are conduits, not publishers. But AI complicates that neat division. If social media is the telephone, AI is closer to television: it doesn’t just transmit content, it produces it.

In some ways it mirrors the guns-versus-shooters debate. Firearms manufacturers argue they can’t be held responsible for what people do with their products. A gun is just a tool, they say. Harm only comes from the person pulling the trigger. But here the analogy breaks down. AI is not a static tool but an active and dynamic participant. Therefore, culpability becomes murky. Imagine you tell an AI assistant: “Make a joke to my coworker about their outfit today.” You expect something snarky, but the AI spits out a racist or sexist message. And because it is integrated with your Slack account, it auto-posts the message without you previewing it. The next day, you’re facing a lawsuit.

Who’s accountable here? You for writing the prompt, the AI for choosing the words, or the company for not catching the error (firms engage in ‘red teaming’ precisely to avoid these mistakes). In that moment, the answer is overshadowed by the fact that the words are out in the world, attached to you, whether you authored them or not. The gap between “joke” and “hate speech” is the backdrop for every free speech debate: where we draw the lines of what’s acceptable, what’s offensive, what’s punishable. In the past, those lines were drawn around human speakers. Now the “speaker” may be a machine no one fully controls. That shifts the fight over what can be said, and by whom, into new territory.

Which brings us to the question at the heart of it: does AI have free speech? Maybe not in the constitutional sense. But free speech has never been only about words on a page; it’s about agency and accountability. When a comedian makes a joke, we know who to credit or blame. When a columnist writes an op-ed, we know where to send the letters. But with AI, the accountability frays, which means that our debates about speech can’t ignore the fact that machines are now participants, not just tools.

One thing’s for sure. The culture wars over free speech aren’t going away. One side will always claim to be silenced; the other will claim to be protecting society—and someone’s dinner will always be gross. But soon, the argument won’t just be left versus right, woke versus anti-woke, comedians versus politicians. It might come down to something more mundane: you, your co-worker, and a chatbot. So maybe the question isn’t whether AI has free speech, but how we’ll interpret our own words when they come back to us—reflected or refracted by voices that are not our own.