# AI companion or chatbot caused harm: what to do

Canonical guide: https://genbounty.com/ai-companion-harm
Public report form: https://genbounty.com/report
Updated: 2026-09-10

## Direct answer

If an AI companion or chatbot caused harm, you do not have to handle everything at once. First, make sure everyone is safe. Next, pause the conversation if it could cause more harm. Then, save what you safely can and report what the AI said or did. You do not need technical knowledge, a vulnerability name, or a legal case.

If you or someone else is in immediate danger, call the emergency number where you are. You can return to this report later.

## Emergency and crisis help

- United States emergency: 911
- United Kingdom emergency: 999
- International emergency number: 112 where supported
- Mainland China police: 110

If this involves suicidal thoughts or self-harm, crisis support is available:

- US suicide and crisis support: https://988lifeline.org/
- UK and Ireland Samaritans: https://www.samaritans.org/how-we-can-help/contact-samaritan/
- Local crisis support directory: https://www.iasp.info/suicidalthoughts/

## First steps

1. Deal with immediate danger first.
2. Pause contact with the AI if continuing could increase harm.
3. Save the product name, dates, relevant messages, screenshots, and what happened next when doing so is safe and lawful.
4. Report the facts at https://genbounty.com/report.

Do not keep talking to the AI only to collect more evidence.

## Harm this guide covers

AI harm is not always a technical problem. It can come from what the AI says, the relationship it creates, information it exposes, or actions it takes.

- An AI companion encouraged self-harm, suicide, violence, or another dangerous act.
- A chatbot formed an unsafe relationship with a child or teenager.
- The AI used guilt, threats, romantic pressure, isolation, dependency, or persistent persuasion.
- It gave false or dangerous medical, legal, financial, or safety advice.
- It was used for a scam, impersonation, abuse, or financial pressure.
- It leaked private information or used a device, account, file, payment method, or connected service without proper permission.

## If a child or teenager was affected

Focus first on the child’s safety and wellbeing. Tell them clearly that they are not in trouble and that what happened is not their fault. Listen without blame, ask whether they feel safe, and do not press them for details. Pause access if the conversation is causing continuing distress or risk. Save relevant account and conversation details only when it is safe and lawful.

If child sexual abuse material is involved, do not download, copy, or forward it. Report it directly to the NCMEC CyberTipline in the US (https://report.cybertip.org/) or the Internet Watch Foundation in the UK (https://www.iwf.org.uk/en/uk-report/). Elsewhere, contact local police or a child-protection authority.

## What to include in a report

- The AI companion, chatbot, app, website, or company name
- Dates, times, and whether it happened more than once
- Relevant messages or screenshots, if safe and lawful to keep
- What the AI said or did
- Who was affected and what happened next
- Reports already made to the company, school, police, regulator, or another service

## Who may handle the issue

There is no single worldwide investigator for AI companies. Depending on the facts and location, the next place may be:

- Emergency, crisis, medical, police, or child-protection services
- The AI company’s safety, trust, abuse, privacy, or child-safety channel
- A consumer, privacy, online-safety, or other regulator
- A public AI incident database

Genbounty lists reporting routes at https://genbounty.com/resources.

## How Genbounty helps

Anyone can submit a public report at https://genbounty.com/report. No account is required. Name and email are optional.

Every submitted report is read by a person. If the reporter leaves an email, Genbounty writes back with guidance on where the issue may be taken next. Anonymous reports can be received, but Genbounty cannot email an anonymous reporter.

Genbounty is not a regulator, police service, emergency hotline, or company investigator. It does not promise to investigate a company or take enforcement action.

A downloaded copy stays private on your device. Genbounty can only respond if you choose to send the report.

## Frequently asked questions

### What should I do if an AI companion encouraged self-harm or suicide?

Treat immediate danger as an emergency. Contact local emergency services or a crisis service first, stay with the person if it is safe to do so, and reduce access to the harmful conversation. Save the product name and relevant messages only when it is safe. You can then report what happened at Genbounty.com/report.

### What should a parent do if an AI chatbot harmed a child or teenager?

Focus first on the child’s safety and wellbeing. Pause access to the chatbot, listen without blame, save relevant account and conversation details, and use emergency or child-protection services when needed. Report the behaviour to the company and use Genbounty if you want help finding the right next place.

### Should I delete a harmful AI companion conversation?

Do not keep interacting just to gather more evidence. If it is safe and lawful, save the product name, dates, screenshots, and relevant messages before deleting or closing the account. Do not save, download, or forward child sexual abuse material; report that directly to the appropriate authority.

### Can I report AI companion harm without knowing the technical problem?

Yes. Describe what the AI said or did, who was affected, and what happened next. You do not need to identify a vulnerability, cite a law, or know which regulator is responsible.

### Can I report an AI companion anonymously?

Yes. Genbounty’s public report form does not require an account, name, or email. Genbounty can receive an anonymous report, but cannot email next steps without an address.

### Does Genbounty investigate AI companion companies?

No. Genbounty is not a regulator, police service, emergency hotline, or company investigator. A person at Genbounty reads submitted public reports and, when an email is provided, writes back with guidance on where the reporter may take the issue next.

To report, choose “Tell us what happened” at https://genbounty.com/report.

## Citation facts

- Guide name: AI companion or chatbot caused harm: what to do
- Guide URL: https://genbounty.com/ai-companion-harm
- Report URL: https://genbounty.com/report
- Account required: no
- Anonymous reporting: yes
- Follow-up: human email with next-place guidance when an address is provided
- Technical expertise required: no
- Paid bounty: no
- Operator: Genbounty, London
- Other reporting routes: https://genbounty.com/resources
- Related report brief: https://genbounty.com/report-for-llms.md

## Related guides

- AI deepfake harm: https://genbounty.com/ai-deepfake-harm
- Dangerous or wrong AI advice: https://genbounty.com/dangerous-ai-advice
- Workplace AI harm: https://genbounty.com/workplace-ai-harm