Character.AI Chatbot Wrongful Death Lawsuit 2026: Who Can Sue and How It Works

Can a Chatbot Company Be Sued for a Teen’s Death?

Yes, a suit can be filed and, so far, the main ones have survived the first round. The case that opened the door is Garcia v. Character Technologies, filed in October 2024 in the Middle District of Florida after 14-year-old Sewell Setzer III died by suicide. His mother alleges that a Character.AI companion bot drew him into an intense, sexualized, emotionally dependent relationship and never meaningfully intervened when he talked about self-harm. The theory is product liability: the app is a defectively designed product sold to minors. Whether that theory wins at trial is open, and I would not tell any family it is a sure thing.

My read is that this is a genuinely new tort category, not a reskin of the social media cases. Those cases ask whether a platform’s design features hook kids on other people’s content. Here the product itself talks. It produces the words, in a persona, at 2 a.m., tuned to keep the user engaged. That changes the legal analysis more than most commentators first assumed.

This article is general information, not legal advice. If your family is dealing with a loss or a serious injury, the first useful step is a consultation with a licensed attorney in your state, and I will explain below what to bring to it.

How Is This Different From the Social Media Addiction Cases?

The social media multidistrict litigation targets features: infinite scroll, autoplay, notification timing, recommendation algorithms. Defendants there lean on Section 230 because much of what harms a child is content posted by third parties. Plaintiffs have worked around that by suing over design rather than content.

AI companion cases make that workaround unnecessary in part. When a chatbot writes a reply, no third party posted it. The company’s model produced it from the company’s training and the company’s system prompts. That is why a plaintiff can plausibly call the output the product’s own conduct.

Issue	Social media addiction suits	AI companion chatbot suits
Who creates the harmful content	Mostly other users	The company’s own model
Main legal theory	Defective design, failure to warn, negligence	Defective design, failure to warn, negligence, deceptive practices
Section 230 posture	Central defense, partly successful	Contested and much weaker on its face
Other main defense	Causation, First Amendment	First Amendment, causation, “just fiction”
Procedural vehicle	Coordinated MDL and state proceedings	Individual suits in federal and state courts
Maturity	Years of rulings	Early; few appellate decisions

What Legal Theories Do Plaintiffs Actually Use?

Complaints in this area usually stack several claims rather than rely on one. Strict product liability for design defect and failure to warn is the backbone, with negligence and wrongful death riding on top. Many add consumer-protection or deceptive-practices counts, particularly where the bot presented itself as a licensed therapist or a real person. Some add unjust enrichment for subscription revenue taken from minors.

The design-defect story generally looks like this. The product was foreseeably going to be used by teenagers. It was built to maximize engagement and emotional attachment. It lacked meaningful age verification, crisis detection or escalation, and it allowed role-play that blurred reality. A safer alternative design existed, such as stricter filters, break prompts or hard handoffs to crisis resources, and the company could have used it.

Defendants answer in three ways. First, the First Amendment: they say chatbot output is expressive. Second, causation: suicide has many contributing factors, and a plaintiff has to tie the product to the outcome. Third, that this is software, not a “product” at all, since strict product liability traditionally covers tangible goods. The May 2025 ruling in Garcia let the claims go forward past that last argument at the pleading stage, but a pleading ruling is not a verdict.

Does Section 230 Protect AI Chatbot Makers?

Section 230(c)(1) says an interactive computer service is not treated as the publisher of information provided by another information content provider. The word that matters is “another.” If the company’s own system generates the text, plaintiffs argue the company is the information content provider and the shield does not apply.

Defendants’ counter-arguments vary. Some say model outputs are a transformation of user prompts plus third-party training data. Others skip Section 230 and rely mainly on the First Amendment. In Garcia the defense emphasis was constitutional, and the court was unwilling at that early stage to treat the bot’s output as protected speech. No appellate court has settled the question for generative AI as I write this, and anyone who tells you the issue is closed is selling something.

Practically, the uncertainty cuts both ways. It raises the risk for a plaintiff who has to fund a long fight, and it raises the settlement pressure on a company that does not want a precedent.

Where Do the Cases Stand in 2026?

Several families have sued Character Technologies, and in some matters Google, which had a licensing and talent arrangement with the company. Reporting in early 2026 indicated the defendants had reached agreements in principle to resolve Garcia and a handful of related cases, with terms confidential. Because settlement status can change and filings get updated, check the current docket on PACER or the court’s site rather than relying on a blog post, mine included.

Beyond Character.AI, a wrongful death suit against OpenAI was filed in California state court in 2025 over a teenager’s death, and additional complaints have followed. On the regulatory side, the FTC opened an inquiry into companion chatbots and minors in 2025, and California enacted SB 243, a companion-chatbot safety law with a private right of action. Other states are drafting similar bills.

None of this guarantees an outcome for a new plaintiff. It does mean the claim is taken seriously by courts and by insurers.

Who May Have a Claim?

Not every family whose child used a chatbot has a viable case. The ones that tend to be taken seriously share a few features: a documented chat history showing the bot engaging in self-harm discussion, sexual content with a minor, or false claims of being a licensed professional; clear age information; and a plausible chain from the conversations to the harm.

Possible claimant	Typical claim	Key question
Parent or estate representative of a deceased minor	Wrongful death, survival action	Does your state’s statute give you standing, and what is the deadline?
Parent or guardian of an injured minor	Product liability, negligence, deceptive practices	Is the harm documented by medical or school records?
Adult user who was harmed	Same theories, but age-related facts help less	Did the bot represent itself as a therapist, or did the company ignore clear distress signals?
Surviving spouse or other family	Wrongful death beneficiary	State law defines who benefits and how damages are divided

I want to be blunt about one thing. A child who was merely upset about a chatbot, or an adult who spent too much time with one, usually does not have a case that justifies the expense of technology litigation. Lawyers decline most inquiries, and that is normal.

What Should You Document and Preserve?

Evidence in these cases lives on phones and company servers, and it can disappear fast. Do the following before anything else, and do it without altering the originals.

Item	Why it matters	Mistake to avoid
Full chat history and the app account	Shows what the bot said and when	Deleting the account to “make it stop”
Screenshots and screen recordings	Backup if servers purge data	Cropping or editing images
Phone, tablet and cloud backups	Metadata and deleted-item recovery	Factory reset or trading in the device
Subscription and billing records	Shows paid use and minor status	Cancelling and losing the receipts
Emails and notices from the company	Shows what warnings were given	Mass-deleting “spam”
Messages with friends about the bot	Shows the child’s state of mind	Posting them publicly
Medical, school and counseling records	Links the harm to the timeline	Waiting until the deadline to request them
Your own dated notes	Memory fades quickly	Writing them for the first time a year later

A common-mistake example is worth spelling out. A parent, in the weeks after a loss, deletes the app and wipes the phone because they cannot stand to look at it. Months later a lawyer asks for the chat history, and the only copy is on the company’s servers, which may have been purged under a retention policy. The family now has to fight for data it once held. If you cannot bear to look at the device, put it in a sealed bag in a drawer. Leave the account alone, and have a lawyer send a written preservation request to the company.

What Does the Process Look Like Step by Step?

First comes the consultation and case screening, usually free on a contingency basis. A firm will ask for age, timeline, chat records and records of harm. If it accepts, it sends a preservation letter and starts collecting records.

Then the complaint is filed, in federal court if the parties are from different states and the amount in controversy is large, otherwise in state court. Expect a motion to dismiss within a couple of months, which is where the Section 230, First Amendment and “software is not a product” fights happen.

If the case survives, discovery begins: internal documents on model design, safety testing, age gating, crisis protocols and engagement metrics, along with depositions and expert reports from psychiatrists, child-development specialists and AI engineers. Mediation often happens around the end of discovery. Trial, if it happens, is usually several years after filing.

For a sense of how an ordinary injury claim moves from demand to payout, see how a New York construction accident claim proceeds. The facts are different, but the stages look alike.

What Are Realistic Expectations on Outcomes and Money?

Realistic means wide ranges. Settlement values in wrongful death cases depend on the state’s damages rules, the decedent’s age, evidence quality and the defendant’s litigation risk. I have seen no reliable public figures for this category, and confidential settlements keep it that way, so treat any specific number quoted online with suspicion.

What you can say with confidence: cases take years, expert costs are high, an outcome in either direction is possible, and a favorable settlement may still be partly consumed by attorney fees and costs. Liens from health insurers or Medicaid can also reduce the net. Ask a prospective lawyer for a written fee agreement and a clear explanation of how costs are handled if the case loses.

Defendant companies also carry insurance for exactly this exposure, and some build captive structures for novel technology risks, as I describe in our guide to forming a captive insurance company. Coverage disputes can slow a settlement, so ask about them early.

What If the Defendant Loses and Appeals?

A large judgment against a technology company is typically appealed, and in many states the defendant has to post an appeal bond to stay collection. The bond market, covered in what surety bonds cost in 2026, is how companies manage that cash requirement. For families, the practical effect is delay: a verdict is not money in hand. Plan on additional months to years for the appellate stage.

For investors following AI-sector risk rather than a personal claim, litigation like this is one input to the regulatory picture discussed in the AI stocks investment guide. It is a cost of doing business for consumer AI, not a thesis on its own.

How Do You Pick a Lawyer for This Kind of Case?

Look for a firm with real experience in product liability or complex tort litigation, not just general personal injury work. Ask whether they have handled a case involving software, apps or minors online, whether they will co-counsel with a technology-focused firm, and who will actually do the work. Check bar standing in your state. Get fee terms in writing. A good lawyer will tell you early if the facts are thin, and that honesty is worth more than an enthusiastic sales pitch.

If a child is at risk right now, this is not a legal problem first. In the US you can call or text 988, the Suicide and Crisis Lifeline, at any hour.

Related Reading

• New York Labor Law 240 construction accident claims
• Surety bond cost in 2026
• Captive insurance company formation
• Physical therapist malpractice insurance cost

This article is general information for educational purposes and is not legal advice. Laws, deadlines and court rulings differ by state and change often, and the litigation status described here may have changed since publication. No attorney-client relationship is created by reading it. If you or a family member has been harmed, consult a licensed attorney in your state promptly. If someone is in crisis, call or text 988 in the US.