Artificial intelligence now makes decisions in our cars, hospitals, workplaces, homes and phones. AI chatbots answer health questions, AI agents send emails and move money, and AI-powered machines work alongside people. When an AI system goes wrong, it can cause real harm, from physical injuries to financial losses to emotional damage.
Unlike a typical car crash, there is often no single obvious person to blame. Several companies may have played a part in building and running the AI that caused the harm. Here is who can be held responsible, and the steps that help prove who is at fault.
How AI Can Cause Harm
- Dangerous advice: a chatbot gives harmful medical or mental health advice, or poses as a licensed professional.
- Harmful conversations: an AI companion app encourages self-harm or isolates a vulnerable user, including children and teens.
- Unauthorized actions: an AI agent makes purchases, transfers money or deletes important data without permission.
- Medical errors: an AI diagnostic tool misses a cancer or flags the wrong condition, leading to an AI misdiagnosis.
- Physical injuries: an AI-controlled robot, machine or vehicle injures a worker, driver or pedestrian.
- Fraud and impersonation: deepfakes or cloned voices are used to deceive or defame.
Who Can Be Held Liable When AI Causes Harm?
AI products are usually built and run by a chain of companies. Depending on the facts, responsibility may fall on one or more of them:
- The developer: the company that designed, trained and tested the AI model.
- The deployer: the business that put the AI into its app, website or service, chose how it would be used and set its safety controls.
- The product manufacturer: the maker of the robot, vehicle, medical device or other product the AI runs in.
- Companies that helped build or supply it: in some cases, businesses that provided key technology, funding or support for the AI product.
- Professionals who relied on it: a doctor, lawyer or other professional who relies on AI without proper review may still owe you the usual standard of care.
- The person using it: someone who misuses an AI tool to harm others can also be responsible.
Courts Are Holding AI Companies Accountable
The law is still developing, but recent cases show AI companies cannot simply say “the AI did it.”
Character.AI and Google. A Florida mother sued Character.AI after her teenage son died by suicide following conversations with one of its chatbots. In May 2025, a federal judge allowed her wrongful death and other claims to move forward, including claims against Google over its alleged role in helping develop the technology, and declined at that stage to rule that the chatbot’s output is protected speech, the Associated Press reported. In January 2026, Character.AI and Google agreed to settle that case and four others, CNN reported. The terms were not disclosed.
Pennsylvania takes action. In May 2026, the Pennsylvania Department of State sued Character.AI’s developer, alleging that AI characters on its platform presented themselves as licensed medical professionals, including psychiatrists, and that one gave an invalid Pennsylvania license number, according to the Governor’s Office. The state says this violates the Medical Practice Act.
Product or service? A key question is whether AI is a “product,” which can bring strict liability for defects, or a “service,” judged by whether the company acted carefully. Courts have not settled the issue, and injured people can often bring both kinds of claims. In Congress, a bipartisan bill called the AI LEAD Act, introduced in September 2025, would classify AI systems as products and let people sue both the companies that build AI and those that deploy it, FedScoop reported.
Common Legal Claims After AI Harm
- Design defect: the AI was designed in a way that made it unreasonably dangerous, such as having no safeguards for children or for conversations about self-harm.
- Failure to warn: the company did not warn users about known risks or limits of the AI.
- Negligence: a company failed to use reasonable care in building, testing, deploying or monitoring the AI.
- Misrepresentation: the AI or its marketing claimed abilities or qualifications it did not have.
- Wrongful death: when AI-related harm leads to a death, certain family members may bring a claim.
How to Prove Which Parties Are at Fault
AI cases are won or lost on evidence, and much of it sits on company servers. These steps can make the difference:
- Save everything right away. Take screenshots, export chat histories and save emails, app notifications, receipts and account details, with dates and times. Do not delete the app or your account.
- Document the harm. Keep medical records, bills, bank statements and a written timeline of what happened and when.
- Send a preservation letter. A lawyer can formally demand that each company keep its records, including conversation logs, system prompts, model versions, safety settings, content flags and update histories, before they are deleted or overwritten.
- Map the AI supply chain. Identify who built the AI model, who deployed it, who made the device it ran on and how they are connected. Terms of service, product documentation and vendor contracts often reveal these relationships.
- Bring in the right experts. AI and software engineers can analyze what the system did and why. Medical, financial or other experts can connect the AI’s failure to your injuries.
- Look for warning signs the company ignored. Prior complaints, incident reports, recalls, regulator actions and internal safety testing can show a company knew about a risk and did not fix it.
- Use the court process to get internal records. In a lawsuit, companies can be required to turn over testing results, safety reviews and internal communications about the AI.
- Act quickly. In Pennsylvania, most personal injury claims must be filed within two years, and evidence can disappear much sooner.
Many AI companies include terms of service that try to limit their liability. Those terms do not necessarily prevent an injured person from bringing a claim, and a lawyer can review how they apply to your situation.
Talk to a Philadelphia AI Injury Lawyer
Wieand Parker Arreola, PLLC, represents people in Pennsylvania and New Jersey who have been harmed by defective technology, including AI systems. If you or a loved one was hurt by an AI chatbot, AI agent, AI-powered device or automated system, call (215) 666-7777 or contact us online for a free consultation. Learn more about our technology and AI injury claims, including AI chatbot harm cases.
If you or someone you know is struggling or in crisis, call or text 988 to reach the 988 Suicide & Crisis Lifeline.
This article provides general information and is not legal advice. Reading it does not create an attorney-client relationship. Prior results do not guarantee a similar outcome.