Requiring business entities to disclose the use of artificial intelligence in certain consumer interactions; establishing the right of consumers to human review in high-impact decisions; and providing for enforcement by Attorney General.
HB1857, the Artificial Intelligence Transparency in Services Act, would require for-profit business entities to disclose when artificial intelligence is being used in a consumer interaction. The disclosure must be clear and conspicuous at the beginning of the interaction, provided in plain language, and delivered orally or in writing in a way that is reasonably accessible to people with disabilities or limited English proficiency. If a consumer asks, and a human representative is reasonably available, the business must provide timely access to that person.
The bill also creates a specific right to human review when AI is involved in a “high-impact decision,” defined to include decisions affecting legal rights, employment, housing, credit, education, health care, or access to government benefits. In those cases, the business must tell the consumer that human review is available, begin that review within 14 days of the request, and finish it within 28 days. The Attorney General would be authorized to enforce the law through civil actions under Pennsylvania’s Unfair Trade Practices and Consumer Protection Law, with penalties of up to $2,500 per violation, while preserving other legal remedies.
HB1857 would add new disclosure and review obligations for businesses that use AI in consumer-facing services, effectively regulating how companies in Pennsylvania may deploy automated systems in customer interactions. It would not ban AI use, but it would require notice, accessibility accommodations, and a human-review process for decisions with major personal consequences. The bill would also expand the Attorney General’s enforcement role under the state consumer protection framework and could expose covered businesses to civil penalties and litigation risk.
No committee transcript or vote record was provided, so there is no documented debate or recorded support/opposition to gauge legislative sentiment. Based on the bill’s structure, it appears framed as a consumer-protection and transparency measure rather than a restriction on AI itself, suggesting an intent to balance innovation with oversight and accountability. The absence of votes or hearing remarks means the public or legislative reaction cannot be reliably characterized from the available record.
The main points of potential contention are likely to be the scope of the disclosure duty, the operational burden on businesses, and the feasibility of providing timely human review. Businesses may object to compliance costs, ambiguity around what counts as an AI-assisted consumer interaction, and the 14- and 28-day review deadlines. Consumer advocates, by contrast, would likely support the bill’s transparency requirements and the right to human review, especially for decisions affecting employment, housing, credit, health care, education, and public benefits.