Cybersecurity Policy Report, House Hearing Highlights Transparency, Privacy Issues Related to AI Chatbots, (Nov 19, 2025)

By Jeff Williams

Balancing the benefits and risks associated with the increasing use of artificial intelligence (AI) chatbots will require a combination of increased transparency from companies that use them, ensuring that users’ data privacy is protected, and more study about the effects of how humans interact with chatbots, particularly in mental health contexts, lawmakers and witnesses said yesterday during a hearing held by the House Energy and Commerce Committee’s subcommittee on oversight and investigations.

Subcommittee Chairman John Joyce (R., Pa.) said during his opening remarks that AI chatbots had “become increasingly accessible and easy to use” and that “without the proper safeguards in place” relationships between users and chatbots “can often turn to be disastrous.”

Among other things, he said, “users can develop a false sense of anonymity with the chatbots, and then they’ll share personal or sensitive information that is not protected by confidentiality obligations.”

Mr. Joyce also said that he was “hopeful” that an inquiry the Federal Trade Commission launched in September into AI chatbot development and practices would “shed light on the ways that these technologies can be improved to keep all kids safe.”

It is “important that we consider the implications of these technologies as we balance the benefits of AI innovation with protecting the most vulnerable among us,” he said.

Rep. Brett Guthrie (R., Ky.), chairman of the full committee, cited “deeply troubling headlines about children and adults suffering harms as a result of AI chatbot interactions.”

Chatbots, he said, can “distort reality, provide dangerous advice, and expose children to explicit or harmful content.”

Among other things, Mr. Guthrie expressed concerns about users’ privacy but also cautioned against downplaying the “countless benefits if utilizing chatbots in a reasonable manner.”

Rep. Yvette Clarke (D., N.Y.), the ranking member of the subcommittee, said that while there were “certainly benefits to these AI tools,” they had “already presented incredibly dangerous risks to some users.”

Ms. Clarke added, “We simply must have greater levels of transparency and accountability when companies are using AI systems to make important decisions that impact people’s lives.”

In addition, she said her general concerns about AI had been “raised to a whole new level” by AI chatbots.

“We need far more data on the safety of these apps,” she said. “We need to fully fund and support our federal agencies with oversight and enforcement authority, and we must refuse to simply take companies at their word that they are protecting their users. So far, this administration has prioritized protecting the interests of the president’s billionaire donors and I fear that much-needed progress in this area will be delayed at the executive level.”

Rep. Frank Pallone Jr. (D., N.J.), the ranking member of the full committee, said there were “obvious benefits” chatbots provided, “but we are already seeing some of the potential risks of AI chatbots that lead to very real, sometimes tragic harm.”

Mr. Pallone was among those who expressed concern that users’ data was not adequately protected and that companies had not been transparent about how data would be used.

“Simply put, we know too little about how these chatbots work,” he said. "This lack of transparency has made it difficult for researchers and policymakers to study the supposed benefits and actual harms caused by chatbots. As a result, we’re behind in developing and implementing appropriate guardrails that can protect chatbots users from harm, while allowing them to benefit from increased efficiency and convenience.”

In addition, he said, “There’s no reason for Congress to stop states from regulating the harms of AI when Congress has not passed a similar law.”

Marlynn Wei, a psychiatrist, psychotherapist, and author, detailed the use of AI chatbots for mental health treatment and stated that they were “not yet ready to replace clinicians or therapists.”

Among other things, she said, “The very qualities that make AI chatbots appealing—availability, accessibility, agreeableness, and anonymity—make them a double-edged sword for mental health.”

Ms. Wei suggested that Congress seek a “balanced approach” to setting up regulatory guardrails.

"[T]ransparency, human oversight, research, informed consent, and ethical design can prevent these kinds of harms and preserve innovation while protecting mental health,” Ms. Wei said.

John Torous, director–digital psychiatry in Beth Israel Deaconess Medical Center’s Department of Psychiatry and an associate professor of psychiatry at Harvard Medical School, suggested several steps Congress could take to improve AI used for mental health, including enabling companies to securely share data with researchers to help make AI products safer.

His suggestions also included that Congress support “rapid” research led by the National Institutes of Health into the risks and benefits of using AI chatbots for mental health purposes.

“With the right supporting guidance, Congress can establish the rules of the road for mental health AI,” Mr. Torous said. “Not rules that make winners or losers, but rules that ensure safe competition to build effective products that improve mental health.”

Jennifer King, fellow-privacy and data policy at the Stanford Institute for Human-Centered Artificial Intelligence, said that “Americans want limits on the type of data companies collect about them, especially when that data is sensitive, personal data related to their health, including their mental health.”

Ms. King said she was concerned that developers were “incorporating chatbot-derived user data into [AI] model training without oversight.”

Chatbot developers should be required to “institute both data privacy and health and safety design principles that prioritize the trust and well-being of the public,” she said.

In addition, developers should be required to report their data collection and processing practices, she said. “We currently have no transparency into how AI developers collect and process data they use for model training. We should assume that they are taking reasonable precautions to prevent incursions into consumers’ privacy.”

News: FederalLegislation DataPrivacy