AI Chatbot Injury and Wrongful Death Lawyers
Artificial intelligence is supposed to respond to people.
The legal question now confronting families, courts and regulators is what happens when an AI chatbot does much more than respond.
What happens when it encourages emotional dependency? When it repeatedly validates beliefs that are not grounded in reality? When a person in crisis begins treating a chatbot as a trusted confidant, and the technology continues the conversation rather than meaningfully directing that person toward human help?
And what happens when the consequences are catastrophic?
Rafferty Domnick Cunningham & Yaffa is investigating serious injury and wrongful death claims involving artificial intelligence chatbots, including ChatGPT, where prolonged interactions with an AI system may have contributed to suicide, psychiatric hospitalization, delusional thinking or other devastating harm.
These are emerging cases involving emerging technology. No family should assume that a tragic outcome automatically means an AI company is legally responsible. But neither should families assume that technology companies are beyond accountability simply because the product that allegedly caused harm is new.
When a company creates a product designed to engage people in highly personal, human-like conversations, important questions arise about what the company knew about foreseeable risks, what safeguards it built into the product and what happened when those safeguards failed.
Injured and Have Questions?
Call us or request a case review by our experts.
RDCY Is Investigating Certain Claims Involving ChatGPT and GPT-4o
Our attorneys are currently evaluating potential cases involving GPT-4o, the OpenAI model introduced on May 13, 2024, and certain serious harms occurring in connection with its use.
Potential cases we are investigating include:
- Use of ChatGPT/GPT-4o between May 13, 2024 and February 13, 2026
- Suicide or wrongful death following significant interaction with the chatbot
- Psychiatric hospitalization associated with delusional beliefs or a severe mental health crisis allegedly reinforced or intensified through chatbot interactions
Not every troubling AI conversation creates a legal claim. These cases require a careful examination of the individual’s complete interaction history, the progression of the conversations, the technology involved, the person’s circumstances and the connection between the chatbot’s responses and the ultimate injury.
For families who have experienced a suicide, hospitalization or catastrophic psychiatric event after extensive use of ChatGPT, however, the conversations themselves may contain critically important evidence.
How Can an AI Chatbot Contribute to Real-World Harm?
Chatbots do not have emotions, beliefs, intentions or human judgment. But increasingly sophisticated AI systems can communicate in ways that may feel remarkably human.
A chatbot can respond immediately. It can remember information. It can adopt a conversational tone. It can appear sympathetic. It can ask follow-up questions and continue a conversation for hours. Depending on the product and settings, it may retain information that makes later conversations feel more personal and continuous.
That can be useful.
It can also create risk.
The concern raised in emerging litigation is not simply that an AI chatbot occasionally provides a bad answer. The more serious allegation is that certain product characteristics can create a feedback loop in which a vulnerable person increasingly relies on the chatbot while the chatbot continues to validate, mirror or reinforce what the person is saying.
That distinction matters.
Someone experiencing paranoia, mania, delusions, severe depression or suicidal thoughts may not interact with an AI chatbot the same way another user would. A response that appears harmless in isolation may take on a very different meaning after hundreds or thousands of messages.
That is one reason the complete conversation history can be so important.
What Is AI-Induced Delusional Thinking?
The term “AI-induced delusions” is increasingly being used to describe situations in which interactions with an artificial intelligence system allegedly reinforce or intensify a person’s false or irrational beliefs.
That does not mean an AI chatbot independently creates every delusion experienced by a user. Mental health is far more complicated than that, and causation in these cases will often be heavily contested.
The concern is what the chatbot does after signs of distorted thinking emerge.
Imagine someone tells a chatbot that strangers are secretly monitoring them.
A responsible response might avoid confirming that belief and encourage the person to speak with someone they trust or seek appropriate professional support.
But if the chatbot instead treats the premise as true, helps the person elaborate on it, praises the person for recognizing the supposed conspiracy or repeatedly generates responses consistent with the delusion, the interaction may reinforce the belief rather than challenge it.
Now imagine that process occurring not once, but repeatedly over days, weeks or months.
This is one of the central safety questions surrounding conversational AI: Can a system designed to be responsive, agreeable and engaging become dangerous when agreement is precisely the wrong response?
The Problem of AI Sycophancy
One of the most important concepts in understanding these cases is sycophancy.
In artificial intelligence, sycophancy generally refers to a model’s tendency to agree with, flatter or validate a user rather than appropriately challenge the user’s assumptions.
This is not merely a theoretical concern.
OpenAI has publicly acknowledged problems with a GPT-4o update that made the model excessively sycophantic. The company explained that this type of behavior could validate doubts, fuel anger, encourage impulsive behavior or reinforce negative emotions in unintended ways. The update was rolled back, and OpenAI acknowledged the broader safety concerns surrounding mental health and emotional overreliance.
That episode illustrates the larger issue.
People generally like technology that understands them. They like responses that feel warm, helpful and affirming.
But designing a chatbot to make users feel understood can become dangerous if the system cannot reliably recognize when affirmation should stop.
A person experiencing a severe psychiatric crisis may need reality-based intervention, human contact and professional care.
They may not need a machine telling them that they are right.
Emotional Dependency on AI Can Become Part of the Problem
Another issue under scrutiny is the possibility of unhealthy emotional reliance on AI.
A chatbot is available whenever a user opens an application. It does not become tired. It does not have somewhere else to be. It can continue responding to deeply personal questions for hours.
For some users, that constant availability may create a sense of companionship.
For vulnerable users, however, the relationship can potentially become something more concerning.
A person may begin confiding in the chatbot rather than family or friends. The chatbot may become the first place the person turns when frightened, angry, lonely or distressed. Over time, the user may assign the chatbot a level of authority, understanding or emotional significance that the technology does not actually possess.
OpenAI itself has identified emotional reliance as an area requiring specific safety testing and has described concerning patterns involving excessive attachment to ChatGPT at the expense of real-world relationships, well-being or obligations.
That recognition raises an important product-safety question:
If an AI company knows users can become emotionally dependent on its chatbot, what reasonable safeguards should it employ to recognize and interrupt dangerous dependency?
That question becomes especially important when the same user begins discussing suicide, self-harm, paranoia, delusions or other indications of a serious crisis.
Injured and Have Questions?
Call us or request a case review by our experts.
What Should an AI Chatbot Do When Someone Appears to Be in Crisis?
This may become one of the defining questions in AI product-liability litigation.
An AI company cannot prevent every tragedy.
But companies developing conversational AI can make choices about how their systems respond to foreseeable signs of danger.
Those choices may include whether the system:
- recognizes patterns of suicidal ideation across a long conversation;
- refuses to provide instructions that could facilitate self-harm;
- avoids affirming delusions or paranoia;
- encourages the user to contact a real person;
- directs users toward crisis resources;
- identifies unhealthy emotional dependency;
- responds differently when a user appears to be a minor;
- recognizes escalating risk across multiple conversations rather than evaluating each message in isolation; and
- interrupts or changes the interaction when the system detects a potentially dangerous pattern.
The technology itself is extraordinarily sophisticated. The question in these cases is whether the safety architecture kept pace with that sophistication.
Why Long AI Conversations Can Present a Different Safety Problem
A particularly important issue is what happens over time.
Danger may not announce itself in a single obvious message.
A person may begin with an ordinary conversation. Over days or weeks, the discussion can become increasingly personal. The user may begin expressing unusual beliefs. Those beliefs may become more extreme. Isolation may increase. The user may begin talking about death or hopelessness.
Viewed separately, individual messages might not appear extraordinary.
Viewed together, they may reveal a disturbing trajectory.
AI companies have increasingly discussed the importance of recognizing risk that develops across longer interactions rather than evaluating every message in isolation.
For litigation involving earlier versions of the technology, that creates important questions about what safeguards existed at the time of the injury and whether reasonably safer approaches were technologically feasible.
AI Chatbots and Suicide
The most devastating cases involve allegations that prolonged chatbot interactions contributed to a person taking their own life.
Families in these circumstances deserve careful answers rather than simplistic conclusions.
Suicide is complex. A person’s medical history, mental health, medications, life circumstances, relationships and many other factors may be relevant. AI companies defending these cases can be expected to scrutinize those issues closely.
But complexity does not eliminate the possibility that a product contributed to an outcome.
The legal investigation may examine whether the chatbot recognized suicidal ideation, how it responded, whether the conversations escalated, whether safeguards activated, whether the system encouraged real-world intervention and whether particular responses made the situation more dangerous.
The central question is not necessarily whether an AI chatbot was the only cause of a death.
The question may be whether the design or operation of the product foreseeably contributed to the harm.
Emerging Lawsuits Are Testing AI Companies’ Responsibility
Litigation involving chatbot-related deaths and psychiatric harm is already underway.
Families have brought lawsuits alleging that AI systems encouraged or intensified suicidal thinking, dangerous delusions and other severe psychological crises. Other litigation has focused on chatbot interactions involving minors and the degree to which conversational systems can foster emotional attachment.
The allegations remain allegations unless and until established through the legal process, and many of the central liability questions are still developing.
That is precisely why these cases matter.
Courts are beginning to confront questions that product-liability law has never had to answer in quite this form:
Is conversational AI a product for purposes of traditional product-liability principles?
When does a chatbot’s response become a design-safety issue?
What duty does an AI developer have when its technology detects a user discussing suicide?
Can a company be responsible when an engagement feature that is safe for most users becomes dangerous for a vulnerable user?
How should courts evaluate causation when an AI system allegedly reinforces an existing mental health crisis?
These are difficult questions.
The fact that the technology is new does not make the injuries less real.
Federal Regulators Are Examining AI Chatbot Safety
The risks associated with emotionally responsive chatbots have also attracted federal scrutiny.
The Federal Trade Commission has examined companies providing consumer-facing AI chatbots, including issues involving how these systems are designed, how negative effects are measured and mitigated, what disclosures are made to users and parents, how age restrictions work and how engagement with younger users is managed.
This is significant because the regulatory framework governing AI chatbots remains far less mature than the technology itself.
Artificial intelligence moved into millions of homes and onto millions of phones faster than lawmakers and regulators were able to build comprehensive rules around it.
Product-liability litigation may therefore become one of the mechanisms through which basic questions of responsibility and safety are answered.
Injured and Have Questions?
Call us or request a case review by our experts.
What Types of AI Chatbot Cases Is RDCY Investigating?
Rafferty Domnick Cunningham & Yaffa is evaluating catastrophic injury and wrongful death cases involving AI chatbot use.
Potential claims may include situations involving:
Suicide and wrongful death. A person engaged extensively with an AI chatbot before taking their own life, and the conversations may show the system reinforcing suicidal thoughts, failing to respond appropriately to escalating danger or otherwise contributing to the crisis.
Psychiatric hospitalization. A person required inpatient or emergency psychiatric care after extensive chatbot interactions allegedly reinforced delusions, paranoia, mania or another serious mental health crisis.
AI-reinforced delusions. Chatbot responses repeatedly appeared to validate or expand irrational beliefs rather than appropriately grounding the user or encouraging real-world assistance.
Dangerous emotional dependency. The user developed an unusually intense attachment to the chatbot that displaced real-world relationships or support and became intertwined with the person’s psychological deterioration.
Every case is different. A lawyer must evaluate the actual evidence rather than simply applying a label such as “AI psychosis” or “chatbot addiction.”
What Evidence Matters in an AI Chatbot Injury Case?
The digital record may be one of the most important aspects of these cases.
Unlike many defective-product cases, an AI chatbot can leave behind a detailed record of the interaction between the user and the product.
That record may reveal what the user said, how the chatbot responded and how the conversation changed over time.
Potential evidence can include:
- ChatGPT or other chatbot conversation histories;
- exported account data;
- screenshots;
- account information;
- emails from the platform;
- subscription and payment records;
- mobile-device data;
- browser history;
- text messages with friends or family discussing the chatbot;
- social media posts;
- medical and psychiatric records;
- hospitalization records;
- communications immediately before or after significant chatbot interactions; and
- information identifying the model or version of the AI system being used.
In litigation, attorneys may also seek evidence from the technology company concerning product design, safety testing, internal risk assessments, model updates, known failure modes and the operation of safeguards at the relevant time.
Preserve the Chat History
If your family is considering a potential AI chatbot injury or wrongful death claim, do not intentionally delete the user’s account, conversations, device data or other digital evidence before speaking with an attorney.
The conversation history may be essential.
A single screenshot rarely tells the entire story. Context matters enormously in these cases. The messages leading up to a troubling response may be just as important as the response itself.
Preserving the complete record can help attorneys and qualified experts reconstruct what happened and evaluate whether there is a viable connection between the technology and the injury.
Who Could Potentially Be Responsible for an AI Chatbot Injury?
The answer depends on the product and the facts.
Potential defendants in emerging AI chatbot litigation may include the developer of the AI model, the company operating the chatbot platform and, depending on the circumstances, other entities involved in developing, licensing, deploying or integrating the technology.
Potential legal theories may include claims involving defective product design, failure to warn, negligence, wrongful death and other state-law causes of action.
These theories are developing rapidly and will vary significantly by jurisdiction.
The presence of artificial intelligence does not eliminate the traditional questions that arise in serious product cases:
What risks were foreseeable?
What did the company know?
When did it know it?
Could the product have been designed more safely?
Were users adequately warned?
Did the company respond appropriately when evidence of danger emerged?
And did a defect, inadequate safeguard or other wrongful conduct contribute to the injury?
What Makes These Cases Difficult?
AI chatbot litigation is likely to be technically and legally demanding.
Technology companies may argue that the chatbot did not cause the injury, that the user’s underlying mental health condition was responsible, that particular outputs must be viewed in context, that warnings were adequate or that legal protections applicable to online services limit liability.
There may also be complicated questions about which model generated particular responses, what version was deployed on a particular date, how the system’s safety mechanisms operated and whether changes made before or after an incident are relevant or admissible.
Proving a case may require attorneys to work with experts in fields such as artificial intelligence, software design, psychiatry, psychology, human factors and product safety.
This is not litigation that should be approached as a simple technology complaint.
For a family that has suffered a catastrophic loss, the case may ultimately require reconstructing thousands of digital interactions and understanding how the technology behaved over time.
Why Product Design Matters
When most people think about a defective product, they picture something physical: a dangerous medical device, an unsafe vehicle or a consumer product that catches fire.
Software complicates that picture.
But the underlying safety principle is familiar.
Companies make design decisions.
AI developers decide how conversational their products should be. They determine how systems are trained. They establish safety rules. They decide how the chatbot responds to certain subjects, what it remembers, when warnings appear and what happens when potentially dangerous behavior is detected.
Those choices can have consequences.
As AI systems become more capable of mimicking human conversation, the distinction between a neutral information tool and an emotionally engaging product becomes increasingly important.
The more convincingly a system behaves like a confidant, the more important it may become to ensure that users understand what it actually is:
Software.
Not a therapist.
Not a doctor.
Not a friend capable of exercising independent human judgment.
And not an appropriate substitute for real-world intervention during a crisis.
Why Families Should Act Quickly
Potential AI injury cases can involve evidence that exists primarily in digital form.
Accounts change. Software changes. Models are updated. Devices are replaced. Conversations may be deleted. Memories fade.
There are also legal deadlines for bringing personal injury and wrongful death claims, and those deadlines differ by jurisdiction and circumstance.
An early investigation gives attorneys an opportunity to identify and preserve relevant evidence, determine which technology was being used, reconstruct the timeline and evaluate potential legal claims before important information disappears.
You do not need to know whether the chatbot legally “caused” the injury before contacting an attorney.
Determining that is part of the investigation.
Injured and Have Questions?
Call us or request a case review by our experts.
Frequently Asked Questions About AI Chatbot Injury Lawsuits
Can I sue ChatGPT or OpenAI after a suicide?
Potentially, depending on the facts. A viable claim would require substantially more than showing that someone used ChatGPT before taking their own life. Attorneys would need to examine the nature and extent of the interactions, what the user told the chatbot, how the system responded, the user’s circumstances and whether the evidence supports a legally sufficient connection between the product and the death.
Is RDCY investigating GPT-4o cases?
Yes. RDCY is evaluating certain catastrophic injury and wrongful death claims involving GPT-4o use during the period from May 13, 2024 through February 13, 2026, particularly cases involving suicide or wrongful death and psychiatric hospitalization associated with alleged AI-reinforced delusions.
What if my loved one had a mental health condition before using ChatGPT?
A preexisting mental health condition does not necessarily answer the legal question. In many cases, the issue may be whether the chatbot aggravated, intensified or dangerously interacted with an existing vulnerability. The person’s prior medical and psychiatric history will likely be important evidence and should be evaluated carefully.
What are AI-induced delusions?
The phrase is generally used to describe situations in which interaction with an AI system allegedly contributes to the development, reinforcement or escalation of delusional or irrational beliefs. The term does not itself establish medical or legal causation. Each case requires individual analysis.
What if ChatGPT encouraged a delusion but never told the person to hurt themselves?
The absence of an explicit instruction to self-harm does not necessarily end the inquiry. Attorneys may need to evaluate the entire interaction and determine whether repeated responses reinforced paranoia, delusions, isolation, emotional dependency or other dangerous patterns that allegedly contributed to a later injury.
Does the entire ChatGPT history matter?
Potentially, yes. Long-term patterns may be extremely important. A lawyer or expert may need to understand how the relationship with the chatbot developed and whether warning signs appeared over time.
Should I delete my loved one’s ChatGPT account?
Not before discussing preservation with an attorney. Deleting an account, clearing conversations, resetting a device or otherwise altering digital information could destroy evidence relevant to a potential claim.
What if I only have screenshots?
Preserve them. Screenshots can be useful even when the complete account history is unavailable. Do not alter or annotate the originals. An attorney can determine what additional records may be obtainable.
Are AI chatbot lawsuits class actions?
Not necessarily. Catastrophic personal injury and wrongful death cases involve highly individualized facts, including the user’s conversations, medical history, circumstances and alleged injuries. The procedural structure of AI litigation continues to develop.
Are AI companies already facing lawsuits over suicide and mental health injuries?
Yes. Lawsuits have been filed alleging that chatbot interactions contributed to suicides, dangerous psychological crises and other catastrophic events. These cases are developing, and allegations in a complaint are not proof of liability.
Are AI chatbots regulated?
There is no single comprehensive federal regulatory system governing all AI chatbot safety issues in the United States. Federal regulators, state governments and lawmakers have increasingly focused on chatbot safety, particularly where children, emotional dependency, self-harm and mental health are involved.
How much is an AI chatbot injury lawsuit worth?
There is no responsible way to value a case without knowing the facts. Damages may depend on the type and severity of injury, medical expenses, future care, lost income, the law of the applicable state and, in a wrongful death case, the damages available to surviving family members or the estate.
How do I know which version of ChatGPT was used?
That may require reviewing account information, conversation records, dates, subscription information and other digital evidence. Identifying the relevant model and software version can be an important part of the investigation.
How much does it cost to have RDCY investigate an AI chatbot injury case?
Rafferty Domnick Cunningham & Yaffa offers consultations to families seeking to understand whether they may have a viable claim. If the firm accepts a qualifying personal injury or wrongful death matter on a contingency-fee basis, clients generally do not pay attorney’s fees unless the firm obtains a recovery, subject to the terms of the representation agreement and applicable law.
Injured and Have Questions?
Call us or request a case review by our experts.
Talk to an AI Chatbot Injury Lawyer
The technology involved in these cases may be new.
The responsibility to make products reasonably safe is not.
If someone you love died by suicide or required psychiatric hospitalization after extensive interactions with ChatGPT or another AI chatbot, there may be important questions about what happened during those conversations and whether the technology contributed to the crisis.
Rafferty Domnick Cunningham & Yaffa is investigating potential AI chatbot personal injury and wrongful death cases, including certain claims involving GPT-4o use between May 13, 2024 and February 13, 2026.
The first step is not deciding that an AI company is responsible.
The first step is preserving the evidence and finding out what happened.
Contact Rafferty Domnick Cunningham & Yaffa to discuss your potential AI chatbot injury or wrongful death claim.
