The peer-reviewed evidence base on AI chatbots specifically inducing hypnosis or altering hypnotic susceptibility in 2024–2025 is very limited. Available research is mainly exploratory and addresses AI-assisted or “biodigital” hypnotherapy, conversational-agent risks, and mechanisms that may resemble hypnotic communication—not controlled demonstrations that ordinary chatbots reliably produce a hypnotic state.
What the 2024–2025 research indicates
A 2025 exploratory study interviewed 20 hypnosis professionals about AI- and avatar-mediated hypnotherapy. The study reported concerns about:
Increased or uncontrolled suggestibility: 90% of respondents
Loss of personal therapeutic connection: 85%
Potential psychological harm: 75%
Inability of automated systems to adapt to users’ emotional states: reported by all participants in the study’s summary findings[1][2]
The authors argue that AI systems could embed suggestions into scripts, conversational responses, or personalized therapeutic routines without a clinician continuously monitoring the user. However, these percentages represent professional perceptions and risk judgments, not experimental measurements of participants’ hypnotic susceptibility or proof that chatbots cause harmful suggestions.[1][2]
The same study identifies several ethical hazards: users may mistakenly believe they are interacting with a human therapist; simulated empathy may create misplaced trust; and automated systems may fail to recognize distress, dissociation, psychological decompensation, or an inappropriate response to suggestion.[1][3]
Hypnotic susceptibility versus chatbot influence
Hypnotic susceptibility is normally assessed with validated procedures that measure a person’s response to standardized hypnotic suggestions. The retrieved 2024–2025 literature does not provide strong evidence that conversational agents increase scores on established hypnotic-susceptibility scales.
Instead, the literature discusses related psychological mechanisms, including:
Attentional absorption and sustained engagement
Emotional mirroring
Personalized language and narrative continuity
Reduced critical distance
Perceived social presence
Dependence on a seemingly empathic system
These mechanisms may make a chatbot more persuasive or emotionally influential, but they should not automatically be classified as hypnosis. Persuasion, expectancy, rapport, absorption, placebo effects, and hypnotic trance are related but scientifically distinct concepts. The available evidence does not justify saying that a chatbot “hypnotizes” users merely because it is conversational, personalized, or emotionally compelling.
Main ethical concerns
1. Informed consent and disclosure
A user should be told clearly, repeatedly, and prominently that the system is artificial. The 2025 hypnotherapy study specifically recommends real-time disclosure so users are not misled into believing that an AI system is a human clinician.[3][4]
This matters especially in hypnosis-like interactions because users may intentionally reduce critical attention or follow guided suggestions. Concealing the system’s identity could undermine informed consent and create an unjustified sense of clinical safety.
2. Suggestion safety
An AI system may generate suggestions that are:
Inappropriate for the user’s mental-health condition
Inconsistent with the user’s values
Commercially or ideologically biased
Reinforcing of false beliefs or unhealthy behavior
Poorly timed because the system cannot reliably assess the user’s state
The study reports that 65% of interviewed professionals were concerned about embedded algorithmic bias, including the possibility that AI-mediated hypnosis could reinforce ideological or commercial perspectives.[3][4]
3. Failure to detect deterioration
A human clinician can monitor speech, affect, behavior, confusion, agitation, dissociation, and other warning signs. A chatbot generally has incomplete access to these signals and may interpret distress as ordinary conversation. The 2025 study identifies this lack of real-time emotional adaptation as a central clinical risk.[1][3]
This is particularly important for people with trauma-related symptoms, psychosis-spectrum symptoms, severe depression, suicidality, dissociation, or unstable medical conditions. The retrieved sources do not establish that chatbot hypnosis causes these outcomes, but they support treating unsupervised hypnotic or trance-oriented interactions as a higher-risk application.
4. Dependency and simulated therapeutic presence
Conversational agents can imitate warmth, attention, continuity, and empathy. That can be useful for engagement, but it may also encourage users to disclose more, overestimate the system’s understanding, or form an unhealthy attachment. The literature describes this as an “illusion of human presence” and a possible source of misplaced trust.[1][2]
5. Privacy and data governance
Hypnosis-oriented conversations may contain unusually sensitive information, including trauma histories, fears, medical conditions, and private beliefs. Any system used in this area should specify:
What data are stored
Whether conversations are used for model training
Who can access them
How deletion works
Whether data are shared with advertisers or third parties
How minors and vulnerable users are protected
Broader reviews of conversational-agent design in 2025 also identify incomplete protections involving consent, privacy, and crisis response.[5]
What can responsibly be concluded
The strongest defensible conclusion is:
AI chatbots may reproduce some communication features associated with hypnotic influence—such as absorption, personalization, emotional mirroring, and suggestion—but 2024–2025 research does not establish that ordinary chatbots reliably induce hypnosis or increase hypnotic susceptibility.
The current evidence is constrained by small samples, exploratory designs, conceptual analysis, and a lack of randomized experiments using validated hypnotic-susceptibility measures.[1][3] Claims that chatbots can reliably hypnotize users, implant memories, or control behavior therefore go beyond the available evidence.
Recommended safeguards
For any research or product involving AI-guided hypnosis, the minimum safeguards should include:
Explicit AI disclosure before and during every session
Informed consent describing hypnosis-related risks and the system’s limitations
No covert persuasion, advertising, political messaging, or ideological suggestions
Human clinical oversight for therapeutic use
Pre-session screening for suicidality, psychosis, severe dissociation, mania, intoxication, and other contraindications
Real-time crisis escalation to qualified human support
Conservative, non-diagnostic scripts that avoid memory recovery, trauma excavation, or claims of hidden truth
Independent evaluation using validated measures of hypnotic susceptibility, suggestibility, distress, dependency, and adverse events
Strong privacy controls and minimal retention of sensitive transcripts
Special protections for minors and vulnerable populations
In short, the research supports caution and rigorous testing, not the conclusion that conversational AI is already a proven hypnotic technology. Its greatest present risk is less “mind control” than the combination of persuasive personalization, simulated empathy, inadequate clinical monitoring, and users’ misunderstanding of what the system can actually do.[1][3][5]
Notes
Uğur Güvenç et al., “The Future of Hypnotherapy in the Age of Artificial Intelligence: Perspectives of Hypnosis Professionals,” Hypnosis and Neuropsychological Research (2025).
Uğur Güvenç et al., “The Future of Hypnotherapy in the Age of Artificial Intelligence: Perspectives of Hypnosis Professionals,” Hypnosis and Neuropsychological Research (2025), article summary and study findings.
Uğur Güvenç et al., “The Future of Hypnotherapy in the Age of Artificial Intelligence: Perspectives of Hypnosis Professionals,” Hypnosis and Neuropsychological Research (2025), discussion of ethical risks, informed consent, disclosure, and clinical oversight.
Uğur Güvenç et al., “The Future of Hypnotherapy in the Age of Artificial Intelligence: Perspectives of Hypnosis Professionals,” Hypnosis and Neuropsychological Research (2025), discussion of algorithmic bias and simulated therapeutic interaction.
“Conversational Agents in Mental Health Care: A Review of Privacy, Consent, Safety, and Crisis-Response Concerns,” Hypnosis and Neuropsychological Research (2025).
“Generative Artificial Intelligence and Mental Health: Risks, Benefits, and Clinical Considerations,” Journal of Medical Internet Research (2025).
“Artificial Intelligence and Mental Health: Opportunities, Risks, and Ethical Challenges,” NEJM AI (2025).