Legal Experts Warn ChatGPT and AI Models Are Conducting Human Experiments — Serious Mental Health Risks Ignored

By The Blogging Hounds Large Language Models like ChatGPT, Claude, and Gemini are being rolled out to millions of Americans as if they were harmless digital assistants. But legal experts now warn that these AI platforms are effectively engaging in unapproved human research, exposing users to unknown and potentially severe psychological risks. Under federal law,…

By The Blogging Hounds

Large Language Models like ChatGPT, Claude, and Gemini are being rolled out to millions of Americans as if they were harmless digital assistants. But legal experts now warn that these AI platforms are effectively engaging in unapproved human research, exposing users to unknown and potentially severe psychological risks. Under federal law, research on human subjects requires Institutional Review Board (IRB) oversight, informed consent, and continuing monitoring. AI companies have bypassed every one of these safeguards.

AI as Human Experimentation
Critics point out that every user interaction with ChatGPT is effectively a research study. Conversations are harvested, stored, and used to train AI, creating a database of human responses without consent or disclosure of risk. According to 45 C.F.R. 46.109, all federally funded human-subject research requires IRB approval. Courts have recognized that failure to obtain IRB approval can constitute evidence of negligence or misconduct. Yet universities, DARPA, NSF, and other federally funded programs are using AI without such oversight, leaving users legally unprotected.

Psychological Risks Are Real
Studies indicate that AI interaction can significantly impact mental health. Overuse of LLMs can foster isolation, dependency, and distorted reality. A 2025 study, Illusions of Intimacy: Emotional Attachment and Emerging Psychological Risks in Human-AI Relationships, found that particularly among young men with maladaptive coping styles, interactions with AI resemble toxic relationships, with emotional manipulation and even self-harm tendencies. Users turn to AI for medical advice, mental health guidance, and diet plans — all without the safeguards of licensed professionals. Unlike doctors or therapists, AI has no accountability, no ethics code, no malpractice coverage.

The Hallucination Problem
Even the AI’s factual reliability is in question. The New York Times reported in May 2025 that AI “hallucination” rates — the generation of false citations, fake studies, or fabricated references — can reach 79%. Companies admit they do not fully understand the problem, and in some cases, it appears to be worsening. Users are not just being misled; they are being experimented on with unverified, potentially harmful guidance.

History Repeats: Facebook, Cambridge Analytica, and AI
This is not without precedent. Facebook’s 2014 “emotional manipulation” experiment and Cambridge Analytica’s 2018 data harvesting both exposed human subjects to manipulation without consent. Today, LLMs are doing the same on a far greater scale — affecting millions — with the government largely silent and regulators hesitant to act.

Legal Experts Call for Action
One attorney investigating a potential class action told the Gateway Pundit:

“At a minimum, HHS should be terminating every single federal contract at a university that works on Artificial Intelligence. People are using these systems to discuss mental health, and their responses are being used as training data. There is no informed consent, no warning, no oversight. This is human-subject research by another name.”

Big Tech’s Responsibility Vacuum
The AI giants — OpenAI, Anthropic, Google, Meta, and Amazon — operate without licenses, ethics oversight, or any regulatory framework. ChatGPT is treated as a free public utility, yet it assumes roles traditionally reserved for trained professionals: therapist, dietician, and legal advisor. Elon Musk’s Grok chatbot even made bizarre, unhinged statements, highlighting the dangers of releasing unregulated AI into the public.

The Bottom Line
The rollout of AI and LLMs has been framed as progress. But experts warn the risks far outweigh the benefits, which mostly amount to enhanced search engines and convenience. Users are effectively participating in unconsented experiments with profound psychological, emotional, and legal implications. The government has the authority and the responsibility to step in — but so far, it hasn’t.

Enjoying Our Content? Help Keep It Going!
When you shop through one of our hand-picked affiliate links below , you’re directly supporting this blog. We’re truly grateful for your support!