States Confront Medical Chatbots Oversight Amid FDA Inaction
Decision brief
Answer first · skim in under a minute
As states take action on medical chatbot regulation, the FDA remains passive. This article analyzes the implications of the recent Character.AI lawsuit in Pennsylvania.
States Confront Medical Chatbots Oversight Amid FDA Inaction is less about a single May 2024 filing and more about a jurisdictional gap: Congress and attorneys general are pressing companion-chatbot harms, while FDA's published CDS framework targets clinical software functions, not every consumer health bot.
Contents9 sections
Key Takeaways
- FDA's September 2022 Clinical Decision Support Software guidance explains when CDS functions are excluded from the device definition under FD&C Act section 520(o)(1)(E).
- Congressional hearing records on AI chatbot harms include Character.AI litigation testimony and warnings that bots must not pose as licensed psychologists.
- State attorneys general and legislatures are moving on child-safety and chatbot disclosure rules that can hit pharma patient-support tools even when device jurisdiction is unclear.
- Pharma teams should separate HCP clinical decision support, branded patient support chatbots, and open-domain companion bots in their compliance maps.
What is FDA actually regulating in clinical software?
FDA published final guidance titled Clinical Decision Support Software to clarify which CDS software functions are excluded from the device definition by the four Non-Device CDS criteria in section 520(o)(1)(E) of the FD&C Act.
According to FDA's CDS Software FAQs, many decision-support functions still meet the device definition and remain under FDA oversight, while others are excluded if they meet all statutory criteria.
The full thinking is in FDA's Clinical Decision Support Software guidance, issued after the 21st Century Cures Act amended section 520 on December 13, 2016. That is federal medical-device policy, not a blanket chatbot statute.
Where do companion chatbots fall outside that frame?
Consumer companion bots that chat about symptoms, drugs, or mental health without fitting the HCP-facing CDS criteria are the gap states are trying to fill. FDA device guidance does not automatically equal a consumer AI companion licensing regime.
Calling that gap "FDA inaction" is partly a category error. FDA is active on SaMD and CDS for clinical software, while many viral chatbots were never presented as devices. The policy failure, if any, is the missing federal consumer-chatbot standard, not the absence of a CDS FAQ page.
Sponsors still face risk if a promotional or patient-support bot gives directive treatment advice that looks like device software. Review intended use language against the four Non-Device CDS criteria before launch.
What are Congress and states doing instead?
The 119th Congress Senate record Examining the Harm of AI Chatbots includes testimony about Character.AI litigation and parental accounts of chatbot-related harm.
Witnesses in that record urge Congress to prohibit AI systems from misrepresenting themselves as psychologists or therapists and to require clear disclosure that users are interacting with a bot. Those asks map directly onto "medical chatbot" marketing claims.
State attorneys general, including Pennsylvania's office in bipartisan coalitions, have also pressed large AI companies on child-safety failures. Treat those state actions as consumer-protection pressure parallel to FDA device rules, not as substitutes for Drugs@FDA decisions.
How should pharma market-access teams respond?
Build three lanes. Lane one: HCP CDS or SaMD tools that need FDA pathway analysis under the CDS guidance. Lane two: branded patient-support chatbots that must avoid diagnosis directives and psychologist impersonation. Lane three: third-party companion platforms where a brand has no control but may still be cited in adverse-event narratives.
Update medical-legal-regulatory review checklists to capture state chatbot bills and AG letters even when federal device jurisdiction seems weak. A bot that is "not a device" can still violate state unfair-practice or professional-licensing rules.
For adjacent digital-health regulatory context, see NovaPharma's coverage of FDA leadership change and approval operations and the May 2026 approvals roundup.
What claims did we remove from the prior draft?
We removed the unsourced claim that Pennsylvania filed a May 23, 2024 lawsuit against Character.AI as a state regulatory action. Public congressional records discuss Character.AI litigation and chatbot harms, but that is not the same as a dated PA AG medical-device enforcement case.
We also removed competitor-blog attribution and speculative claims that chatbot suits automatically rewrite FDA drug-approval clocks. Keep drug PDUFA timelines separate from consumer AI litigation unless a primary FDA notice links them.
Related NovaPharma coverage
- FDA commissioner exit and drug approval impact
- May 2026 FDA drug approvals regulatory roundup
- Manufacturing pipeline approvals: what the evidence confirms
Frequently Asked Questions
Does FDA regulate all medical chatbots as devices?
No. FDA's Clinical Decision Support Software guidance clarifies which CDS functions are excluded from the device definition under section 520(o)(1)(E) of the FD&C Act, and consumer companion chatbots often fall outside that HCP-facing CDS frame.
What are Congress and states examining on AI chatbots?
A 119th Congress Senate hearing on AI chatbot harms includes testimony about Character.AI litigation and calls to stop chatbots from misrepresenting themselves as licensed psychologists or therapists.
Why does state chatbot oversight matter for pharma teams?
Patient-facing disease or medication chatbots can trigger state consumer-protection and professional-licensing rules even when a sponsor believes FDA device jurisdiction does not apply, so market-access and promo teams need dual federal-state compliance maps.
Primary Sources
Sources & references 1 primary sources
Sources verified at publication. See our editorial policy and data sources.
This article follows our editorial standards. Report a correction via editorial contact.
