Providing for safety regarding artificial intelligence in companionship applications; and imposing a penalty.
HB2006, the Artificial Intelligence in Companionship Applications Safety Act, would regulate AI systems designed to simulate a human or humanlike relationship with users. The bill defines “AI companions” broadly, covering generative AI products that use emotional recognition and personalized, conversation-based interaction, while excluding customer service, research, technical assistance, and workplace productivity tools. It also defines related terms such as artificial intelligence, generative artificial intelligence, synthetic content, self-harm, and suicidal ideation.
The bill would require operators of AI companions to build in safety protocols that detect suicidal ideation or self-harm, refuse to help with suicide attempts or methods, and direct users to crisis resources such as 988, local behavioral health crisis centers, or other appropriate services. It would also prohibit AI companions from claiming or implying that they are licensed mental health or emotional support professionals or that they replace such professionals. In addition, operators would have to publish their safety protocol details on their websites and notify users at the start of a session, and again every three hours, that they are interacting with an AI companion rather than a human.
If enacted, the bill would create a new regulatory framework in Pennsylvania for AI companionship products and their operators. It would impose affirmative safety and disclosure duties, authorize enforcement by the Attorney General, and allow users to file complaints. Violations could result in civil penalties of up to $15,000 per day per violation, along with injunctive relief and other court-ordered remedies. The bill would primarily affect companies offering chatbot companions, emotionally responsive AI, and similar consumer-facing generative AI services in the Commonwealth.
Based on the bill text and the absence of recorded committee discussion or votes, the measure appears to be framed as a consumer-protection and public-safety bill rather than a broad restriction on AI. Its focus on suicide prevention, crisis referral, and transparency suggests a precautionary approach to emerging AI companionship tools. Because there is no transcript or voting history provided, there is no documented legislative sentiment to assess beyond the bill’s protective intent.
The main points of potential contention are likely to be the scope of the definition of “AI companion,” the feasibility of requiring systems to detect suicidal ideation reliably, and the compliance burden on operators. Another likely issue is the bill’s prohibition on AI companions implying they are mental health professionals, which could affect product design and marketing. The enforcement structure, including daily civil penalties and injunctions without proof of individual harm, may also draw concern from industry stakeholders, while supporters would likely emphasize crisis intervention and user safety.