The US has no law written specifically for personal AI agents as of November 2026. That doesn't mean agents are unregulated. The FTC Act's ban on deceptive practices, the children's privacy rule COPPA, and federal card dispute rights all apply to agents just as they do to any other consumer product, and states are passing their own AI laws. Meanwhile a December 2025 executive order aims to challenge state AI rules that conflict with federal policy.

Is there a federal AI agent law?

No. Congress has not passed a statute addressing AI agents, and the FTC has not issued agent-specific guidance that we could find as of November 2026. Instead, several general laws do the work:

Rule What it covers Relevance to agents
FTC Act, Section 5 Unfair or deceptive practices Misleading claims about what an agent does or how safe it is
COPPA Online data of children under 13 Collecting children's data without verifiable parental consent
Fair Credit Billing Act Credit card billing errors Disputing a charge for goods not accepted or not delivered as agreed
Regulation E Debit and electronic transfers Liability caps for unauthorized debit transactions
FTC Impersonation Rule Impersonating government and businesses Scammers posing as AI vendors (rule exists; we have not quoted its text)

What has the FTC done about AI?

  1. Operation AI Comply: five FTC actions over deceptive AI claims, including DoNotPay and Rytr
  2. FTC finalizes COPPA amendments
  3. FTC 6(b) orders to seven AI chatbot companies on effects on children and teens
  4. FTC reports $15.9 billion in consumer fraud losses for 2025
  • Operation AI Comply (September 2024) brought five actions, including against DoNotPay and Rytr. Then-Chair Lina Khan said: "Using AI tools to trick, mislead, or defraud people is illegal" (FTC). The point for agents: marketing claims about safety and capability have to be true.
  • COPPA amendments (January 2025) require separate parental consent before sharing children's data with third parties and limit retention (FTC). Most personal agents are 18+ anyway; see are AI agents safe for teens?.
  • The 6(b) inquiry (September 2025) asked Alphabet, Character.AI, Instagram, Meta, OpenAI, Snap and xAI how they "measure, test, and monitor potentially negative impacts of this technology on children and teens". It is a study, not an enforcement action.

What protects you if an agent spends your money?

Card protections are the most practical safety net.

Credit card dispute window after the statement (FCBA)
60 days
Max debit loss if reported within 2 business days
$50
Max debit loss if reported later, within 60 days
$500
  • Credit cards: send a written dispute "within 60 calendar days after the charge appeared on your statement"; the issuer must acknowledge within 30 days and resolve within 90 (CFPB).
  • Debit cards: loss is capped at $50 if you report within two business days, up to $500 after that, and possibly unlimited more than 60 days after the statement (CFPB).

The unsettled question is whether a purchase you approved in an agent, but didn't intend, counts as "unauthorized". Don't assume a refund. Our guides on unapproved purchases and who pays when an agent gets it wrong cover the steps. Report fraud at reportfraud.ftc.gov and ic3.gov.

What are the states doing?

States are moving faster than Congress, but few laws target agents directly. Two examples we could confirm:

  • California SB 243, signed 13 October 2025, with most duties effective 1 January 2026, regulates companion chatbots. According to the Future of Privacy Forum, it requires disclosures, extra protections when an operator knows a user is a minor, and gives injured people a private right of action. Whether it reaches task-focused agents depends on how a product is designed and used.
  • The Colorado AI Act (SB 24-205) targets algorithmic discrimination in high-stakes decisions. A law signed 28 August 2025 pushed its start date from 1 February to 30 June 2026 so lawmakers could amend it (Greenberg Traurig). We have not confirmed what happened in the 2026 session, so check current records before relying on it.

Other state laws exist, but we have not researched them for this page. For texting agents, federal TCPA rules also apply; see the TCPA guide on SMS AI Agents.

Will federal policy override state laws?

Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence", was signed on 11 December 2025 (govinfo). It directs the Attorney General to set up an AI Litigation Task Force to challenge state AI laws inconsistent with a "minimally burdensome" national policy, and asks Commerce to identify conflicting state laws. Legal commentary notes it preserves state authority over child safety. An executive order cannot by itself repeal state law, so state rules remain enforceable unless courts or Congress decide otherwise.

What about federal security guidance?

NIST, CISA and other agencies publish guidance, but it is voluntary and aimed mostly at organisations. Its core advice still carries over to personal use: grant agents least privilege, require human approval for consequential actions, and assume prompt injection can't be fully prevented. See what NIST, CISA and the NCSC say.

Bottom line

In the US, your strongest protections against an agent gone wrong are the ones you already have: card dispute rights, the FTC's power over deception, and COPPA for children. Use a credit card or single-use card with your agent, keep the activity log, and act fast if something looks wrong. For comparison, the EU has binding transparency rules; see our EU AI Act guide, and the UK's regulators set out their position in a joint statement covered by Personal Agent UK.