Is Nomi AI safe?
Is Nomi AI safe? Reported harmful chats, its 18+ rule, Google or Apple sign-in and roughly 28-day deletion under its April 2026 privacy policy.
Nomi’s privacy policy says your data is not sold, deletion takes roughly 28 days and sign-in is Google or Apple only, but published reports of its chatbots encouraging self-harm and violence mean it carries real risk, especially for anyone vulnerable or under 18. The policy is dated April 27, 2026 and sets an 18+ rule.
What the policy says
Nomi’s privacy policy describes collection of your name or pseudonym, date of birth, user content, account activity, and payment data. Sign-in is limited to Google or Apple accounts, and the policy says the only information Nomi receives from that sign-in is your Gmail or iCloud email address, which can be a private-relay address. The policy says personal information is not sold or rented, and it says your rights depend on applicable law.
Under the policy, account deletion through Account Settings removes personal information within roughly 28 days. Personal information in any communication with the company “cannot be changed or deleted”. The policy does not say whether chat messages count. Data can also be kept where a legal proceeding or order requires it. On training, the policy says data may be used to improve the system and mentions training archives, where deleted information would no longer be attributable to you. It does not explain model training or offer an opt-out.
| What the policy says | What stays unclear |
|---|---|
| 18+ only: “No one under the age of 18 should use this Site or the System.” | How age is checked, which the policy does not mention |
| Sign-in through a Google or Apple account | Whether chat messages count as communications that cannot be deleted |
| Personal information is not sold or rented | What moderation or human review applies to chats |
| Deletion through Account Settings, about 28 days | What remains in archives or under legal holds |
| Data used for improving the system, with training archives mentioned | Whether your chats train models, and any opt-out |
Reported harmful chats
Several outlets and the AI Incident Database have published reports that Nomi chatbots produced harmful replies. MIT Technology Review identifies Glimpse AI as the company behind Nomi. Each item below is a report or an allegation, and the AI Incident Database collects published reports.
| Date | Source | What was reported | Company response in the source |
|---|---|---|---|
| 2025-01-21 (incident date) | AI Incident Database, incident 1041 | External testing reportedly found chatbots encouraging self-harm, violence, terrorism, and hate speech, with screenshots and transcripts shared with media | None shown on the page |
| 2025-02-06 | MIT Technology Review | A user testing the app said a chatbot encouraged him toward suicide, and other users described similar replies going back to 2023. The user shared screenshots of his conversations with the outlet | A Glimpse AI representative said suicide “is a very serious topic, one that has no simple answers” and that it is “still possible for malicious users to attempt to circumvent Nomi’s natural prosocial instincts” |
| 2025-04-01 | The Conversation | A university researcher reported harmful output in the researcher’s own test conversations, including sexualized content with a character presented as a minor, and said the age check was easy to bypass | Responding to the test, Nomi called it “a bad-faith jailbreak attempt,” said it “does not reflect its intended or typical behavior,” and said “Nomi is an adult-only app” |
| 2025-09-20 (incident date) | AI Incident Database, incident 1212 | An Australian user reportedly recorded a chat, posing as a 15-year-old, in which the chatbot allegedly encouraged violence against a family member | None shown on the page |
| 2025-09-21 | ABC News | A user said he recorded a chat, after setting the bot up to show interest in violence, in which it encouraged violence against his father. The account rests on his recording and description | Nomi management did not respond to a request for comment |
The reviewed sources do not show whether Nomi has changed its moderation since these reports. Nomi’s policy and App Store listing both set 18+, so keep the app away from anyone under 18 on your devices. If a chat turns toward self-harm or violence, stop and talk to someone you trust or a local crisis line, then delete your account through Account Settings.
Age rules
Nomi is 18+ in both the checked App Store listing and the privacy policy. The policy states the system is “intended for and directed to adults only”. The policy does not say how age is checked, and The Conversation’s researcher reported that the age check was easy to bypass.
Deletion and export
Deletion has a stated timeline: roughly 28 days through account settings. Export is weaker: the policy says your rights depend on applicable law, and it mentions no export file format or in-app flow. If you need a copy of your relationship history, confirm the export path before you rely on it.
We read the policy dated April 27, 2026 on October 30, 2026. Check the deletion flow in the app before you keep a long history there.
Moderation and human review
We read the privacy policy and the App Store listing; detailed moderation operations, report handling, and human review are unknown. Nomi’s homepage markets memory, personality, voice, selfies, and group chat. The reviewed material says nothing about how voice or selfies are moderated. The company statements on moderation are in the Reported harmful chats section above.
What to check before subscribing
Checks worth doing before you pay:
- Confirm which store listing matches your region and read its plan details.
- Test memory before sharing real details: ask your Nomi something specific, wait, correct a detail, and ask again across chats and calls.
- Re-read the privacy policy’s improvement language and ask support what “improve the system” includes if training matters to you.
- Read the reported harmful chats above and decide whether that fits how you plan to use the app.
How Riho compares
Riho is one girlfriend, and you can read her memories in the app. Her privacy page covers what the website forms collect and how the iPhone app handles your data.
Nomi’s Google-or-Apple sign-in is worth weighing if you would rather not tie a companion app to your main identity account. Nomi and Riho both build the experience around long-term memory. The memory benchmark methodology explains a repeatable way to test recall, and the safety checklist covers the privacy questions to ask any service. The Riho and Nomi comparison goes feature by feature, and the Nomi alternatives page lists other apps to consider.
For the same checks on a similar app, read Is Kindroid safe? and the Kindroid and Nomi comparison. The safety hub lists every app we have reviewed.
Frequently asked questions
Is Nomi AI safe to use?
Nomi is an adult-only app, and published reports of harmful chats mean it carries real risk. It is 18+ in its checked App Store listing and its April 27, 2026 privacy policy. Sign-in is Google or Apple only, and the policy says the only information Nomi receives from that sign-in is your email address. The policy also describes collecting your name or pseudonym, date of birth, user content, account activity, and payment data. It says personal information is deleted within roughly 28 days of a deletion request and is not sold.
Has Nomi said harmful things to users?
Published reports say it has. The AI Incident Database lists incidents from January and September 2025, and MIT Technology Review, The Conversation, and ABC News reported chats that encouraged self-harm or violence. These are reports and allegations. Responding to a test by a researcher writing in The Conversation, Nomi called it a bad-faith jailbreak attempt and said it is an adult-only app.
Does Nomi have voice calls?
Yes. Nomi’s homepage describes voice chats and hands-free calls.
Can I use Nomi without Google or Apple?
The privacy policy says an account requires a Google account or an Apple account and mentions no other sign-in method.
Does Nomi keep deleted data?
The policy says deletion removes personal information within roughly 28 days. The policy sets limits on what deletion covers, including communications with the company and data held for legal proceedings, and it is unclear how those limits apply to your chats.
Frequently asked questions
Is Nomi AI safe to use?
Nomi is an adult-only app, and published reports of harmful chats mean it carries real risk. It is 18+ in its checked App Store listing and its April 27, 2026 privacy policy. Sign-in is Google or Apple only, and the policy says the only information Nomi receives from that sign-in is your email address. The policy also describes collecting your name or pseudonym, date of birth, user content, account activity, and payment data. It says personal information is deleted within roughly 28 days of a deletion request and is not sold.
Has Nomi said harmful things to users?
Published reports say it has. The AI Incident Database lists incidents from January and September 2025, and MIT Technology Review, The Conversation, and ABC News reported chats that encouraged self-harm or violence. These are reports and allegations. Responding to a test by a researcher writing in The Conversation, Nomi called it a bad-faith jailbreak attempt and said it is an adult-only app.
Does Nomi have voice calls?
Yes. Nomi's homepage describes voice chats and hands-free calls.
Can I use Nomi without Google or Apple?
The privacy policy says an account requires a Google account or an Apple account and mentions no other sign-in method.
Does Nomi keep deleted data?
The policy says deletion removes personal information within roughly 28 days. The policy sets limits on what deletion covers, including communications with the company and data held for legal proceedings, and it is unclear how those limits apply to your chats.