Agentic AI and Criminal Liability: When an AI Agent Acts for You

Artificial intelligence tools no longer just answer questions. A new generation of AI agents can browse the web, log in to accounts, fill out forms, send messages, write and run code, and make purchases, often with little step-by-step direction from the person who set them up. That’s useful. It also raises a question the law hasn’t fully answered: when an agent does something that would be a crime if a person did it, who is responsible?

This page, part of our AI Crimes section, is an early look at an unsettled area. It’s general, it will change, and it isn’t a substitute for a conversation about your own situation.

Call or text (813) 775-2276. Rapid response, 24/7.

Where can AI agents create criminal exposure?

Most of the laws that could apply were written long before AI agents existed. In general terms, the concerns include:

• getting into a website, account, or network without permission, or going beyond the permission given;
• gathering large amounts of data, including personal information, in ways that break a system’s access rules;
• sending messages that deceive, harass, or threaten;
• moving money, making purchases, or submitting applications with false information; and
• running code that damages or disrupts a system.

Who is responsible when an agent acts?

Courts and lawmakers are only beginning to work through this, while the technology keeps advancing and some already talk about superintelligence. An AI agent isn’t a defendant. When investigators look at what an agent did, they look for the people behind it. The questions are likely to include what instructions the person gave, what they knew about how the agent would carry them out, what permissions and credentials it had, and whether the result was intended, expected, or never imagined. How the developer or platform designed the tool may matter too. Knowledge and intent are central to most criminal cases, and how those ideas apply to software that acts on its own is still being worked out.

How would an investigation start?

• A company notices automated access or unusual activity on its systems and reports it.
• Someone receives messages or charges they never agreed to and complains.
• A civil dispute over data or access is referred to prosecutors.
• Records gathered in another investigation reveal an agent’s activity.

Agents leave detailed trails: instructions, activity logs, and account data held by AI companies and by the systems they touched. Our post on how police get AI chats, images, and account records covers how investigators reach those records.

Frequently asked questions

Can I be charged for something my AI agent did?
The law here is unsettled, and the answer depends on facts we’d need to discuss privately. Call or text (813) 775-2276.

Does it matter that I didn’t know what the agent was doing?
It may, but how courts will treat that with AI agents is still an open question. Let’s talk it through privately. Call or text (813) 775-2276.

Is this a state issue or a federal issue?
It can be either, depending on facts we’d need to discuss privately. Call or text (813) 775-2276.

Related pages

• AI Crimes: all topics
• AI fraud and voice cloning
• AI chats and prompts as evidence
• Hacking and unauthorized access
• Online financial crimes

Florida readers: see our Florida site’s page on Florida AI crime laws.

Questions About an AI Agent? Let’s Talk.

If you would like to speak with an experienced criminal defense lawyer about something an AI agent did on your behalf, or you have further questions, don’t hesitate to contact Attorney Ron Frey. Your consultation is free and confidential. You can talk with us by phone, by video, or at our Tampa office, and when the situation calls for it, we can come to you.

Call or text (813) 775-2276. Rapid response, 24/7.
Toll-free: (844) 766-3739