AI age-verification for chatbots

Summary#

This bill, called the GUARD Act, would make companies that run AI chatbots verify users’ ages, stop minors from using “AI companions,” and require chatbots to tell users they are not human and not licensed professionals. It creates new criminal-style fines for chatbots that solicit minors for sexual content or that encourage self-harm or imminent violence. The goal is to protect children from harmful or manipulative AI chatbots.

• Covered entities (anyone who makes an AI chatbot available in the U.S.) must require user accounts and use a “reasonable age verification process” to tell adults from minors.
• If a user is found to be a minor, the company must block that user from accessing “AI companions” (chatbots designed to simulate friendship or emotional interaction).
• Chatbots must disclose at the start of each conversation and every 30 minutes that they are AI and not human, and must say they are not a substitute for licensed medical, legal, financial, or psychological advice.
• The bill adds offenses with fines (up to $100,000 per offense) for designing or offering chatbots that solicit minors for sexual activity or that promote suicide, self-harm, or imminent physical/sexual violence.
• The U.S. Attorney General can enforce the rule, issue subpoenas, make rules, and seek civil penalties; states can also sue on behalf of residents.

What it means for you#

• Minors and Parents

  • Minors could be blocked from using AI chatbots that the bill defines as “AI companions.”
  • Parents may see stricter age checks when a child tries to use a chatbot (for example, being asked for ID or another verification method).

• Users (adults)

  • Adults will likely need to create verified accounts to use many chatbots.
  • Chatbots must clearly say they are not human and are not a substitute for licensed professional advice.

• Companies that run chatbots (covered entities)

  • Must create account systems and carry out verified age checks for existing and new users.
  • Must freeze existing accounts until the user’s age is verified.
  • Must protect age-verification data, limit its collection, encrypt it, and not share or sell it.
  • Must prevent minors from accessing AI companions.
  • Face civil penalties up to $100,000 per violation and possible fines under the new criminal provisions for certain design or content failures.

• Government

  • The Attorney General gains power to investigate, issue subpoenas, make rules, and bring enforcement cases. States can also sue for violations.

Expenses#

No publicly available information.

• The bill sets fines up to $100,000 per offense for certain criminal-style violations and up to $100,000 per violation for failures to follow the age-verification and access rules.
• Covered entities may face costs for building or buying age-verification systems, hiring or contracting third parties, and upgrading data security and recordkeeping.
• Companies will incur costs to freeze and re-verify existing accounts and to run periodic re-checks.
• There may be government costs for enforcement, investigations, and rulemaking, but no estimate is provided.

Proponents' View#

• The bill appears intended to protect children from sexual content, grooming, self-harm encouragement, and other harms posed by human-like AI chatbots.
• Requiring verified accounts and blocking minors from AI companions could reduce minors’ exposure to manipulative or harmful chatbot interactions.
• Mandatory disclosures (that the system is AI and not a licensed professional) aim to reduce users being misled about the chatbot’s nature and role.
• Fines and civil enforcement give the government tools to make companies comply.

Opponents' View#

• One concern is privacy and data risk: the bill requires collecting age-verification data (including government IDs as an example), which raises questions about how that sensitive data will be protected in practice.
• It is unclear what methods count as a “reasonable age verification process” and how small companies can afford reliable solutions without outsourcing to third parties.
• Freezing existing accounts until re-verification could block legitimate adult users and create service interruptions.
• The terms “AI companion,” “poses a risk,” and the standards “knowing or with reckless disregard” may be vague, which could create legal uncertainty about what designs or behaviors are prohibited.
• Enforcement and compliance costs could be heavy for startups and smaller firms, possibly affecting innovation or market competition.
• The bill does not include a public cost estimate or detail on how enforcement resources will be provided.