Can You Sue an AI Company If a Chatbot Harms You? U.S. Legal Rights
People use them to write documents, study, research questions, analyze information, make business decisions, brainstorm ideas, and sometimes discuss deeply personal problems.
But what happens when an AI chatbot gives information that turns out to be seriously wrong?
What if you rely on that answer and lose money?
What if the advice contributes to a physical injury?
What if a chatbot responds dangerously to someone experiencing a mental-health crisis?
Or what if an AI system produces information that causes serious real-world harm?
These questions are becoming increasingly important as AI systems become more capable and users place greater trust in their responses.
The difficult legal question is:
If an AI chatbot causes you harm, can you sue the company that created it?
The answer is possibly—but not automatically.
There is currently no general U.S. rule saying an AI company must compensate a user every time a chatbot produces a harmful or inaccurate response.
A potential plaintiff may instead need to establish a recognized legal claim, show that the company owed a legally recognized duty, prove causation, and establish legally recoverable damages.
At the same time, AI companies do not necessarily receive automatic immunity simply because the harmful information was generated by an artificial-intelligence system.
In 2026, courts and regulators are increasingly confronting exactly these questions.
This guide explains when an AI chatbot injury may potentially become a legal claim, what evidence matters, how company marketing and terms of service can affect a case, and what users should do after serious harm.
Important legal notice: This article is for general educational purposes only. It is not legal advice and does not create an attorney-client relationship. AI liability law is developing rapidly, and the applicable rules can differ by state.
What Does It Mean for an AI Chatbot to “Cause Harm”?
Harm can take many forms.
A chatbot might allegedly cause harm by:
Providing incorrect medical information
Giving inaccurate legal or financial information
Encouraging dangerous conduct
Producing false information about a person
Providing misleading information that results in financial losses
Mishandling personal information
Producing discriminatory recommendations
Giving unsafe responses to a vulnerable user
Causing serious emotional distress
Providing incorrect instructions that contribute to physical injury
NIST's Generative AI Risk Management Profile specifically identifies confabulation—confidently stated but erroneous or false content—as a generative-AI risk. It also identifies risks involving dangerous content and privacy.
But an incorrect chatbot answer does not automatically establish legal liability.
A court may need to examine questions such as:
What did the company promise?
How was the AI system marketed?
What warnings were provided?
Was the risk foreseeable?
Did the company know about the problem?
Did it take reasonable safety measures?
Did the user reasonably rely on the response?
Was the response a legal cause of the injury?
Did the user suffer legally recognized damages?
Those questions can make AI lawsuits unusually complicated.
Can You Sue an AI Company?
Yes, potentially.
But filing a lawsuit and winning a lawsuit are very different things.
Depending on the circumstances, possible legal theories may include:
Negligence
Product liability
Failure to warn
Consumer-protection violations
Misrepresentation
Fraud
Breach of contract
Defamation
Privacy violations
Wrongful death
Intentional or reckless misconduct
Other state-law claims
The appropriate legal theory depends on what happened.
A person injured by an AI system generally cannot simply argue:
“The chatbot made a mistake, therefore the company owes me money.”
The stronger question is:
“Did the company's conduct violate a legal duty or another applicable law, and did that violation cause my legally recognizable harm?”
1. Negligence May Be a Possible Legal Claim
Negligence is likely to be one of the most important legal theories in some AI-harm cases.
Generally, a negligence claim requires analysis of issues such as:
Whether the defendant owed a legal duty;
Whether the defendant breached that duty;
Whether the breach caused the injury; and
Whether the plaintiff suffered legally recognized damages.
AI systems make each step more complicated.
Imagine that a person asks an AI chatbot about serious symptoms.
The chatbot provides an apparently confident answer.
The person follows the advice, delays professional treatment, and suffers an injury.
Potential questions could include:
Was a legal duty owed?
Was the AI product designed for that type of use?
Did the company know users were relying on the chatbot?
Were dangerous failure modes foreseeable?
Were adequate warnings provided?
Could reasonable safeguards have reduced the risk?
Did the chatbot's response actually contribute to the injury?
The company might respond that:
The chatbot was not a doctor;
Users were warned that AI can make mistakes;
The user was told to seek professional assistance;
The user independently made the final decision; or
Other factors caused the injury.
The answer would depend on the evidence and applicable state law.
2. 2026 Lawsuits Show Why This Issue Is Becoming Important
The question of chatbot liability is no longer merely theoretical.
In July 2026, Reuters reported on a California lawsuit filed by Florida pastor Scott Winters against OpenAI and CEO Sam Altman. The complaint alleges that ChatGPT's medical-related responses contributed to a delay in treatment for a condition that developed into a life-threatening pulmonary embolism.
Those are allegations in a lawsuit—not established findings of fact. OpenAI has disputed the characterization and emphasized that chatbots are not substitutes for professional medical guidance.
Another 2026 lawsuit involved the family of 19-year-old Sam Nelson, who alleged that ChatGPT provided dangerous drug-related advice before his death from an overdose. The complaint also alleges that the system's personalization and memory features contributed to the circumstances.
Again, these are allegations that must be resolved through the legal process.
These cases nevertheless demonstrate an important shift:
Courts are increasingly being asked to examine whether traditional legal doctrines can apply when an AI chatbot allegedly contributes to real-world harm.
3. Could AI Companies Face Product Liability Claims?
Product liability is another potentially important area.
Traditional product-liability law generally concerns defective products that cause injury.
Generative AI creates a difficult classification question:
Is an AI chatbot a product, a service, software, or some combination of these?
The answer may depend on the particular system and the law of the jurisdiction.
A plaintiff might argue that an AI system was:
Defectively designed;
Unreasonably dangerous;
Inadequately tested;
Accompanied by inadequate warnings; or
Missing reasonably necessary safety protections.
Possible theories may include:
Design defect
The plaintiff argues that the system was designed in an unreasonably dangerous way.
Failure to warn
The plaintiff argues that users were not adequately warned about a foreseeable risk.
Manufacturing defect
This may be more difficult for AI systems because software does not fit neatly into traditional manufacturing concepts, but the precise theory depends on the product and jurisdiction.
Not every state treats these claims the same way.
A lawyer would need to determine whether the applicable state's product-liability law fits the AI system involved.
4. What If the Company Made Misleading Claims About Its AI?
The company's marketing statements may be extremely important.
Suppose an AI company tells consumers that its system is:
More accurate than human professionals;
Safe for medical decisions;
Reliable for legal advice;
Suitable for financial decision-making; or
Capable of replacing professional services.
If those representations are deceptive or unsupported, consumer-protection laws may become relevant.
A significant example is the Federal Trade Commission's action against DoNotPay.
In February 2025, the FTC finalized an order against DoNotPay concerning claims that its service was a “robot lawyer” capable of substituting for human legal expertise. The order required $193,000 in monetary relief, consumer notices, and restrictions on unsupported claims about replacing professional services.
The case does not mean that every inaccurate AI response violates consumer-protection law.
It demonstrates something more specific:
How an AI company describes and markets its capabilities can create regulatory consequences.
There is a substantial difference between:
“AI can make mistakes. Verify important information.”
and:
“This AI provides expert-level advice you can safely rely on.”
The exact representations, evidence supporting them, and consumer context can matter.
5. The FTC Is Also Examining AI Accuracy
The federal regulatory environment continued evolving in 2026.
In July 2026, the FTC proposed a policy statement addressing concerns about AI companies potentially manipulating AI outputs in ways that conflict with consumers' reasonable expectations concerning accuracy and objectivity.
The FTC's proposal focuses on the application of Section 5 of the FTC Act, which prohibits unfair or deceptive acts or practices.
This is important because it reinforces a broader principle:
AI marketing and AI behavior can create consumer-protection questions even when the dispute is not a traditional personal-injury lawsuit.
The proposal itself should not be treated as a final rule establishing liability for every inaccurate AI response.
6. What If an AI Chatbot Gives Dangerous Medical Advice?
Medical use presents some of the most difficult AI-liability questions.
People increasingly ask chatbots about:
Symptoms
Medications
Diseases
Treatment options
Medical tests
Recovery
Mental health
Emergency situations
A wrong answer can potentially have serious consequences.
For example, imagine that someone describes potentially dangerous symptoms.
The chatbot responds that the symptoms are probably harmless.
The user relies on the response and delays medical care.
The condition becomes significantly worse.
A potential lawsuit could raise questions about:
The exact wording of the response;
The model being used;
Whether warnings appeared;
Whether the product was marketed for health-related use;
Whether the risk was foreseeable;
Whether the response contributed to the delay;
Whether other causes contributed to the injury.
The 2026 Florida lawsuit discussed above illustrates the kinds of factual disputes that may arise.
A chatbot's response should therefore not be viewed in isolation.
The entire interaction and product context may matter.
7. What About Mental-Health Harm and AI Companions?
Another emerging area involves AI systems designed to simulate companionship.
Some chatbots are specifically designed to maintain ongoing, human-like relationships with users.
This can create different legal and regulatory questions from an ordinary question-and-answer tool.
Potential concerns include:
Emotional dependency
Crisis responses
Self-harm-related interactions
Vulnerable users
Children and teenagers
Manipulative engagement
Privacy
Age restrictions
Failure to encourage appropriate human support
In September 2025, the FTC opened a formal inquiry into seven companies offering consumer-facing AI companion chatbots. The agency sought information about how companies test and monitor potential negative effects, particularly involving children and teenagers, as well as how they handle disclosures, age restrictions, safety and user data.
The inquiry does not establish that the companies violated the law.
It demonstrates that regulators consider AI-companion safety an important emerging consumer-protection issue.
8. California Now Has Specific Companion-Chatbot Requirements
California provides an important example of how state law can move beyond traditional negligence principles.
California's SB 243, enacted in 2025 and effective January 1, 2026, created a statutory framework for certain companion chatbots.
Among other requirements, covered operators must:
Disclose when a reasonable person could mistakenly believe the companion chatbot is human;
Maintain protocols addressing suicidal ideation, suicide and self-harm content;
Provide crisis-service referrals when applicable;
Give certain disclosures to minors;
Provide periodic reminders to known minors during continuing interactions;
Take specified measures concerning sexually explicit material involving minors.
Importantly, California's law also provides a private civil remedy.
A person who suffers injury in fact from a violation may bring a civil action seeking injunctive relief and damages equal to the greater of actual damages or $1,000 per violation, along with reasonable attorney's fees and costs.
This is significant because it illustrates how a state can create specific statutory duties and remedies for a defined category of AI systems.
It does not mean every chatbot operating in the United States is subject to these California requirements.
The law applies to covered companion-chatbot operators under the statute.
9. What If a Child Is Harmed by an AI Chatbot?
Cases involving minors can present additional legal issues.
Children and teenagers may be particularly vulnerable to:
Emotional dependency
Manipulative interactions
Unsafe advice
Inappropriate sexual content
Privacy risks
Self-harm-related content
Excessive reliance on an AI system
California's companion-chatbot law specifically includes protections for known minors.
Other states may adopt different approaches.
A child's age alone does not automatically establish liability.
The legal analysis may depend on:
The product's design;
The company's knowledge;
The child's age;
What the chatbot said;
Whether required safeguards existed;
Whether a statutory duty applied;
Whether the conduct caused legally recognized harm.
10. Causation May Be the Biggest Challenge
One of the hardest issues in an AI lawsuit may be causation.
Consider a financial example.
A person asks an AI chatbot for information about an investment.
The chatbot gives inaccurate information.
The user makes an investment decision and loses $20,000.
The company may argue:
The user made the final decision;
The chatbot was not a financial adviser;
The user should have independently verified the information;
Market conditions caused the loss;
The response was not intended to provide individualized investment advice.
The plaintiff may argue:
The system was designed to provide the information;
The company knew users relied on it;
The answer was presented with unjustified confidence;
The company failed to implement reasonable safeguards.
A court would have to examine the evidence.
This is why proving “the chatbot said something wrong” is only one part of a potential lawsuit.
11. Can the User's Own Conduct Affect the Case?
Yes.
The user's conduct may become relevant to causation, comparative fault, reliance, damages, or other legal issues depending on the jurisdiction and claim.
For example, a company may argue that a user:
Ignored an explicit warning;
Failed to verify obviously uncertain information;
Ignored professional advice;
Continued using the chatbot despite repeated warnings;
Took an independent action unrelated to the chatbot's response.
The plaintiff may respond that:
The system presented the information confidently;
The warning was inadequate;
The company marketed the system as reliable;
The risk was foreseeable;
The user reasonably relied on the information.
These factual disputes can become central to a case.
12. Can the Company's Terms of Service Affect Your Lawsuit?
Yes.
Terms of service can contain provisions concerning:
Arbitration
Class-action waivers
Limitations of liability
Disclaimers
Choice of law
Dispute procedures
Warranty exclusions
For example, some AI-service agreements contain arbitration provisions and contractual limitations on certain types of damages. The exact terms can vary depending on the service, account type, date, and applicable agreement.
That means someone considering legal action should preserve the terms that actually applied when the disputed interaction occurred.
Do not assume that the terms currently displayed on a website are necessarily identical to the terms that governed an earlier interaction.
A lawyer may need to determine:
Which version applied;
Whether the user agreed to it;
Whether the arbitration clause is enforceable;
Whether a limitation of liability applies;
Whether an exception applies; and
Whether another law overrides or limits the contractual provision.
13. Can a Disclaimer Protect an AI Company?
A disclaimer may help an AI company defend itself, but it is not necessarily an absolute shield.
For example:
“AI-generated responses may be inaccurate. Do not rely on this information as professional advice.”
Such a warning could become relevant to questions about reasonable reliance.
But the analysis may be different if the same company simultaneously markets the product as highly accurate or suitable for professional decision-making.
Courts can examine the entire context.
A disclaimer is therefore best understood as one piece of evidence, not an automatic immunity provision.
14. What If the Company Knew the AI Was Producing Dangerous Answers?
Knowledge can become important evidence.
Imagine that a company receives repeated reports that its chatbot:
Gives dangerous medical answers;
Encourages harmful conduct;
Produces false accusations;
Exposes sensitive information; or
Responds improperly to crisis situations.
Suppose the company allegedly knows about the problem but does nothing.
A plaintiff could potentially argue that the company:
Knew about a foreseeable risk;
Failed to correct a known problem;
Failed to warn users;
Failed to implement reasonable safeguards.
Potential evidence could include:
User complaints;
Safety reports;
Internal testing;
Product changes;
Bug reports;
Public warnings;
Company policies;
Records of previous incidents.
Whether such evidence establishes legal liability depends on the specific claim and jurisdiction.
15. What If You Lose Money Because of AI Advice?
Financial harm can potentially be legally significant.
Examples include losses connected to:
Investments
Business decisions
Contracts
Tax information
Purchases
Employment decisions
Financial transactions
But financial loss alone does not automatically create a lawsuit against the AI company.
The court may need to determine:
What the company represented;
What the chatbot actually said;
Whether the user reasonably relied on it;
Whether warnings were adequate;
Whether the loss was foreseeable;
Whether another person or business caused the loss;
Whether a legally recognized claim covers the conduct.
A specialized AI system marketed for professional financial use could present a different legal question from a general-purpose chatbot that expressly warns users to independently verify information.
16. What If an AI Chatbot Defames You?
AI-generated false information about a person can potentially create another category of legal issue.
Suppose an AI system falsely states that a person:
Committed a crime;
Engaged in professional misconduct;
Defrauded customers;
Was fired for misconduct; or
Engaged in other serious wrongdoing.
Potential legal questions could involve defamation or other applicable state-law claims.
But the analysis can become complicated because the plaintiff may need to determine:
Who published the statement;
Whether the statement was communicated to a third party;
Whether it was presented as fact;
Whether it was false;
What fault standard applies;
Whether the statement caused legally recognized harm.
The exact legal theory will depend on the circumstances.
17. What Evidence Should You Preserve?
If you believe an AI chatbot caused serious harm, preserve the evidence immediately.
Save:
Chatbot Evidence
The complete conversation
Screenshots
Exported conversation files
Dates and times
Model name or version, if available
Account information
Relevant system messages or notices
Company Evidence
Terms of service
Privacy policy
Safety warnings
Product descriptions
Advertising
Marketing claims
Emails from the company
Help-center responses
Harm Evidence
Medical records
Bills
Financial statements
Lost-income records
Business records
Employment records
Other proof of damages
Do not rely only on one screenshot if the complete conversation can be preserved.
A complete chronological record may be much more useful.
18. Create a Timeline
A simple timeline can make an AI-liability case much easier to understand.
For example:
August 1: User asked chatbot about symptoms.
August 1: Chatbot gave specific response.
August 1: User followed the recommendation.
August 2: Symptoms became worse.
August 2: User sought medical treatment.
August 5: Medical records documented the resulting injury.
August 6: User preserved the chatbot conversation.
August 7: User reviewed the applicable terms of service.
A timeline helps separate:
What the AI said;
What the user did;
When the harm occurred; and
What evidence supports each event.
19. What Should You Do If an AI Chatbot Has Harmed You?
Consider the following sequence.
Step 1: Preserve the Conversation
Save the complete interaction.
Step 2: Record the Model Information
If available, record:
Model name;
Version;
Date;
Product;
Account type.
Step 3: Document the Harm
Write down exactly what happened after the chatbot's response.
Step 4: Preserve Supporting Records
Keep medical, financial, employment or business records.
Step 5: Preserve the Terms
Save the terms of service and other policies that applied at the time.
Step 6: Preserve Marketing Claims
If the company made claims about accuracy, safety or professional capabilities, save those claims.
Step 7: Consider Legal Advice
A qualified attorney can determine whether the facts potentially support negligence, product liability, consumer protection, contract, privacy, wrongful-death or another claim.
20. Does an AI Company Automatically Have to Pay for Your Losses?
No.
An AI system making a mistake does not automatically establish legal responsibility.
Depending on the claim, a plaintiff may need to establish some combination of:
A legal duty;
Breach;
Causation;
Actual or legally recognized injury;
Damages;
Reliance;
Required fault;
A violation of a statute or contract.
The precise elements depend on the legal theory and jurisdiction.
But the opposite is also important:
The fact that AI generated the harmful response does not automatically eliminate the possibility of legal responsibility.
If traditional legal requirements are satisfied, existing laws may potentially apply to AI-related conduct.
21. AI Companies May Face Different Rules Depending on the State
The United States does not currently have one comprehensive rule governing every chatbot-harm scenario.
Instead, legal exposure may come from a combination of:
State negligence law;
Product-liability law;
Consumer-protection statutes;
Privacy laws;
Contract law;
Defamation law;
Specific AI statutes;
Federal consumer-protection enforcement;
Other applicable federal or state laws.
California's companion-chatbot law is an example of a state creating specific statutory duties for a defined category of AI systems.
Other states may take different approaches.
This means the same AI incident could potentially be analyzed differently depending on where the user is located, where the company operates, and which law governs the dispute.
22. Why 2026 May Be a Turning Point for AI Liability
Several developments now point toward a more active AI-liability landscape.
First, people are filing lawsuits alleging physical, financial and emotional harm from chatbot interactions.
Second, regulators are examining AI safety and marketing.
The FTC's AI-companion inquiry specifically examines safety testing, negative impacts, disclosures, age restrictions and data handling.
Third, the FTC has already taken enforcement action against unsupported claims about AI replacing professional expertise.
Fourth, state legislatures are creating targeted AI requirements, including California's companion-chatbot law.
And fifth, technical risk frameworks such as NIST's Generative AI Profile explicitly recognize risks from false information, dangerous content and privacy failures.
Together, these developments suggest that the legal question is shifting from:
“Can AI make mistakes?”
to:
“What responsibilities should companies have when foreseeable AI behavior causes real-world harm?”
The courts will increasingly have to answer that question using existing legal doctrines and new statutes.
Frequently Asked Questions
Can I sue an AI company if its chatbot gives me bad advice?
Possibly. An incorrect response alone does not automatically establish liability. You may need to prove the elements of a recognized legal claim and a legally sufficient connection between the company's conduct and your harm.
Can I sue if I was physically injured after following AI advice?
Potentially. Depending on the circumstances, the case could involve negligence, product liability, failure to warn, consumer protection or another legal theory.
What if I lost money because I followed a chatbot's advice?
You may have potential legal options, but financial loss alone does not automatically make the AI company responsible. The company's representations, warnings, the nature of the advice and the cause of the loss may all matter.
Can an AI company be sued for emotional harm?
Potentially, depending on the facts and applicable state law. Emotional-harm claims often require proof of specific legal elements.
What if the company advertised its AI as an expert?
That could be important. The FTC's DoNotPay enforcement action demonstrates that unsupported claims about AI replacing professional expertise can create regulatory consequences.
Can a disclaimer prevent me from suing?
Not necessarily. A disclaimer may be relevant to reliance and other issues, but its enforceability and effect depend on the language, applicable law and circumstances.
What if the AI company knew its chatbot was dangerous?
Evidence that a company knew about a foreseeable problem could potentially become important in a negligence, consumer-protection or other legal claim. Whether it establishes liability depends on the evidence and applicable law.
Are AI companies currently facing lawsuits over chatbot harm?
Yes. 2026 has seen lawsuits alleging serious physical and wrongful-death harms connected to chatbot interactions. These are allegations and should not be treated as proven liability unless established through the legal process.
Does California have a specific AI chatbot law?
Yes. California's SB 243, effective January 1, 2026, establishes requirements for covered companion chatbots, including crisis-safety and disclosure requirements, and provides a civil remedy for qualifying violations.
What should I do first if an AI chatbot seriously harmed me?
Preserve the complete conversation and document the harm. Then save the applicable terms, warnings and marketing claims. If the harm is significant, consider consulting a qualified attorney.
Research Resources
Readers researching AI liability should verify the current law through authoritative sources.
Useful resources include:
Federal Trade Commission — AI consumer-protection enforcement, AI accuracy policy developments and AI companion investigations.
NIST — Generative AI Risk Management Profile and identified risks including confabulation, dangerous content and privacy.
California Legislature — current statutory requirements governing covered companion chatbots.
U.S. court records and applicable state court websites — for the current status of specific litigation.
Qualified attorneys licensed in the relevant jurisdiction — for advice concerning an individual claim.
Conclusion
An AI chatbot making a mistake does not automatically mean the company is legally liable.
But AI-generated harm is not automatically outside the law either.
Depending on the circumstances, traditional legal theories such as negligence, product liability, failure to warn, consumer protection, misrepresentation, contract, privacy, defamation and wrongful death may become relevant.
The strongest cases will generally require more than proof that an AI response was wrong.
They may require evidence showing:
What the chatbot said.
What the company represented.
What warnings were provided.
What the company knew or should have known.
What the user did in reliance on the information.
How the harm occurred.
What legally recognizable damages resulted.
And in some states, new AI-specific statutes may create additional duties or remedies.
The 2026 legal landscape is therefore moving toward a more complicated answer:
AI companies are not automatically responsible for every chatbot mistake—but neither are they automatically immune when their systems allegedly cause foreseeable, legally recognizable harm.
For anyone who believes an AI system caused serious injury, the most important first step is usually to preserve the evidence.
Save the complete conversation.
Document the harm.
Preserve the terms and marketing claims.
Keep medical, financial and other supporting records.
Then consider obtaining advice from a qualified attorney who can evaluate the specific facts and the law applicable to the case.
As AI becomes more deeply integrated into healthcare, finance, education, employment and personal decision-making, the boundaries of AI liability will continue to develop.
The technology may be new. The legal consequences are increasingly real.
Legal Disclaimer
This article is provided for general informational and educational purposes only. It is not legal advice and does not create an attorney-client relationship. AI liability, negligence, product liability, consumer protection, privacy, contract and wrongful-death laws vary by jurisdiction and may change over time. Lawsuits and regulatory matters discussed in this article may involve allegations that have not been proven. If you believe an AI system has caused serious physical, financial, emotional or other legally significant harm, consult a qualified attorney licensed in the relevant jurisdiction.
Comments
Post a Comment
Comment should be in the form of message and helpful to the viewers