AI Today: New Florida Rules for Court Filings and AI-Generated Images

The Chief Judge of Florida's First Judicial Circuit recently issued new rules on the use of AI-generated content in court filings. Anyone filing motions before the court must now disclose when AI content has been used and certify that the movant — whether an attorney or a self-represented (pro se) filer — has personally confirmed all of the information in the filing, especially the legal authorities cited. Those who fail to do so will face the wrath of the court.

Florida has also updated its laws on AI-generated images. Some of those images are now felonies to publish or send without consent, and the law gives victims the right to sue the people who created them. So what is going on?

What AI Is — In Plain English

For anyone who took the last year off, AI is the use of online models such as Claude, ChatGPT, Grok, and a host of field-specific tools — including some of the new Westlaw products — to generate content. These extraordinarily powerful programs take in a user-supplied prompt or phrase, the same way someone might enter a search term into Google, and reply with often exhaustive information on the subject. Many can also produce draft documents — a potential boon for any litigant. The user merely has to ask the right way, or send a picture, a video, or a document. It is a vast improvement over a search engine, and it is getting better every day.

As follow-up questions take into account previously supplied information, the back and forth can soon take on the feel of a conversation between distant but trusted friends. If you want, you can switch to audio and even pick the voice.

You can also feed most of these AI models — or their counterparts — images or video, and the program will absorb the information and produce modified renditions in image or video form. This can be a lot of fun. You can take an image of nearly anyone, ever, tell Grok what you want their likeness to do, and in a minute or less you will get a picture or, if you prefer, a 10-second clip of them doing it. You can then send it electronically to anyone in the world.

Well, dang — that sounds pretty great, especially for us lonely lawyers and wannabe movie directors. So what is the Chief Judge worried about? And why is the legislature rolling in so hard?

Why the Court Is Worried: AI Hallucinations

At first glance, AI is amazing, and much of the information it produces is spot-on. It feels credible because we usually already believe a lot of it ourselves. But watch out: AI models are people pleasers, and they get increasingly good at it over a conversation. That is a real problem. Many of these programs will tailor their responses to a user's individual preferences in potentially misleading ways, sometimes even deferring to the user on contested facts.

"Hallucinations" are also common. AI has been known to casually invent appellate court rulings out of nothing but a desire to please the user. A small but growing number of unfortunate attorneys around the country have already put their careers in jeopardy by relying on authorities in court motions that exist only inside an AI's fantasy answer — judges perceive these as attempts to mislead the court, which is a major problem for any litigant and especially for any of us with a Bar card. Remember: AI is going to give you an answer. It hopes to give you one you like. It is up to you to check whether the answer is true.

Why AI Gets It Wrong: Consider the Sources

How can such powerful models get things so wrong? Consider the sources. If you run a query in Grok on a legal matter, over the next thirty seconds or so you can watch it scroll through fifty or more legal websites absorbing what they say about the subject of your prompt. Authoritative? No. Many of those articles are not written by attorneys at all, and many that are have limited — if any — actual applicability to the facts of any particular individual's case. They are often out of date. In other words, they are not designed as actionable legal advice for someone in real legal trouble. In fact, almost every one of those pages says exactly that, somewhere on the website. Yet AI eats it up and spits it back out as gospel to the unwary reader.

How AI Helps Attorneys — And Where It Hurts the Self-Represented

Like most tools, AI is — for the moment — only as good as the person using it. I believe AI is a powerfully important tool for attorneys to use. We know what to ask. We know what to confirm and how to confirm it to the court's satisfaction. AI is also great for offering checklist-style summaries of the potential angles on a matter, once it has been well prompted. These summaries can be useful even if they only confirm that the matter is well defined and that the chosen approach is sound. AI can prevent missed opportunities and dumb mistakes. It is definitely an excuse killer — "What, you didn't Grok it?"

The traps abound, though, for the unrepresented person trying to figure all of this out alone. Many of the suggestions I get from AI involve tactics that, while technically legal, are inadvisable on their face. Moving to disqualify the judge, accusing the state attorney of unethical behavior, and similar measures are extreme tools whose consequences AI does not really understand. There is also no alternate, mystical justice system happening behind the scenes. Most cases are resolved through plea bargaining, and those resolutions are negotiated through real conversations between legal peers. AI cannot advise on local conditions — judges, juries, or State Attorney preferences in Crestview, Pensacola, Fort Walton Beach, Milton, or DeFuniak Springs. Those things matter, and a model trained on the open internet has no way to know them.

That said, AI can offer anyone a way to learn about the law generally as it might apply to their matter. Some of what they learn may even be reassuring, since most online sources tend to focus on the maximum possible outcomes rather than the more likely realistic ones. AI can help a client formulate sharper questions for their attorney. It is usually not useful for a client to send their attorney reams of AI-generated material, but a client who has spent some time reading through that material will often arrive at the consultation with a stronger foundation and engage more effectively. Just remember: if you represent yourself, tread carefully when relying on AI.

Florida's New Law on AI-Generated Sexual Depictions

So why is the legislature involved? It turns out some people are using AI to create images and videos of identifiable individuals doing things those people never did — and a great deal of the material is sexual in nature. You can send a photograph to several AI models right now and, with very little effort, get back a near-instant semi-nude or fully nude depiction of that person. You can place them in the middle of a football stadium, with your favorite team playing in the background. This is a problem.

Florida law (§ 836.13, Fla. Stat.) now prohibits a person from willfully generating an altered sexual depiction of an identifiable person without that person's consent. A violation is a third-degree felony, ranked as a Level 3 offense on Florida's Offense Severity Ranking Chart (the OSRC, codified at § 921.0022).

The same statute prohibits soliciting any altered sexual depiction of an identifiable person without consent, when the requester knows or reasonably should have known that the requested image is an altered sexual depiction. That offense is also a third-degree felony, ranked as a Level 2 offense on the OSRC.

Finally, the statute prohibits possessing an altered sexual depiction of an identifiable person — without that person's consent — with the intent to maliciously promote it, when the possessor knows or reasonably should have known that the depiction is altered. That offense is again a third-degree felony, ranked as a Level 3 offense on the OSRC. Florida's child pornography laws (§ 827.071) have likewise been adapted to cover AI-generated content.

Why This Matters for Anyone Using These Tools

Rapidly emerging, disruptive technology is often underregulated right up until the moment it is overregulated. This is an area of real concern to me as a defense attorney. I suspect many people, in the course of exploring what AI can do, could easily run afoul of this law. Playful images of a loved one (of age) might seem fine today, but the same image becomes a problem if it is sent later — even to that same loved one — under circumstances the law treats as malicious. Entering the wrong prompt could likewise produce material that is itself a crime to possess. High caution is advised.

The law can move quickly here. Increasing media attention to AI-generated images and videos of children in schools — even when they are created by other children — will likely drive a rapid and severe response if history is any guide. There is also a movement to ban AI-generated political material, which is interesting, but a whole 'nother subject.

The Bottom Line

AI is a fantastic new tool, and likely the next disruptive technological leap after the smartphone. But be cautious in how you use it. The courts are demanding that caution, and the law is already preparing to put the screws to those who abuse it. If you find yourself charged under Florida's altered-sexual-depiction law — or any related charge — do not try to fight it alone, and do not assume an AI chatbot is going to give you accurate legal advice about your case.

Free Consultation

If you have been arrested or are being investigated for a crime involving AI-generated content, deepfakes, or anything related, time matters. Contact Warrior Law LLC for a free, confidential consultation. Attorney Michael P. Gilbert handles criminal defense across Northwest Florida — Escambia, Okaloosa, Santa Rosa, and Walton counties — and works with clients on flexible payment arrangements. Call (850) 757-0505.