New York Advances Bill to Ban AI Chatbots in Licensed Professions, Despite Evidence of Superior Diagnostic Performance
Key Takeaways
- ▸S7263 holds AI companies liable for chatbot responses in 13+ licensed professions, with penalties that cannot be waived by disclaimers
- ▸Large language models have demonstrated superior performance to physicians on diagnostic benchmarks, with Claude 3.5 scoring 30 percentage points higher than junior radiology faculty
- ▸The AMA reports 81% physician adoption of AI tools and opposes blanket bans, viewing AI as a collaborative tool rather than a replacement
Summary
New York State's Senate Bill S7263 has advanced through committee with a unanimous 6-0 vote, proposing to hold AI companies civilly liable for chatbots providing substantive responses in 13+ licensed professions including medicine, law, dentistry, and engineering. The bill, sponsored by Sen. Kristen Gonzalez (D-Queens), would make it illegal for consumers to receive AI-generated advice in these fields, with liability that cannot be waived through disclaimers. However, the legislation faces significant criticism for ignoring peer-reviewed evidence showing AI systems substantially outperforming human professionals on diagnostic tasks.
Recent benchmarking data reveals that every large language model tested on the NEJM Image Challenge—a diagnostic assessment taken by over 60,000 physicians—significantly outperformed doctors across multiple specialties. Claude 3.5 scored 80.4% in neuroradiology compared to first-year radiology fellows at 71.4% and junior faculty at 51.8%. Despite this evidence, and with 81% of physicians now incorporating AI into their practice according to AMA surveys, S7263 seeks to restrict patient access to these tools. The bill reaches the full Senate floor on February 26, 2026, with a vote imminent, even as critics argue it prioritizes workforce protections over public health benefits.
- The bill prioritizes workforce protection over public health despite evidence that medical errors remain the third leading cause of death in the U.S.
- Over 900,000 New Yorkers lack insurance and 92% of low-income legal problems go unaddressed, raising questions about restricting access to AI-powered guidance
Editorial Opinion
S7263 represents a troubling disconnect between legislative intent and available evidence. While framed as consumer protection, the bill appears primarily motivated by workforce concerns—a distinction Senator Gonzalez's own press releases inadvertently reveal. Most concerning is the bill's disregard for compelling performance data: if large language models consistently outdiagnose human physicians on standardized benchmarks, restricting patient access to them raises serious public health questions. The irony is particularly sharp given New York's massive gaps in professional services access; banning AI tools that could democratize expertise seems counterproductive when 92% of low-income legal problems go unaddressed and hundreds of thousands lack insurance.
