What’s Changed: Nomi’s current Privacy Policy (April 27, 2026) and Terms (January 16, 2026) say it collects little and never sells your data, but chats are not end-to-end encrypted and feed anonymized training. Nomi is safe enough for adults who keep identifying details out of their chats. It is for adults only and offers no crisis support.
What is Nomi AI: An AI companion app from Glimpse.ai, Inc. of Maryland where you chat, call and roleplay with characters called Nomis that remember past conversations.
Is Nomi AI safe? Start with what its own documents say deletion means, because they do not agree.
The Complaints Policy says a deleted account’s content is “permanently and irreversibly removed”, while the Privacy Policy keeps anything in its “training or communications archives”.
I want that settled before I type anything personal into an app, and deletion is only one of the questions Nomi’s record raises.
Nomi also has a documented incident record, a Google Play listing rated 17+ despite its 18+ rule, and Terms that let it close any account “for any reason.”
The five-step setup further down keeps your exposure low, and the verdict at the end says who should use Nomi and who should skip it.
Is Nomi AI Safe for Your Private Chats?
Nomi AI is reasonably safe for private chats if you keep identifying details out of them. It collects little, says it never sells data, and strips names before training, but chats are not end-to-end encrypted and some archived data survives deletion.
If you are comparing companion apps on these same points, Candy AI is the one I check Nomi against first.
The Privacy Policy, last updated April 27, 2026, says Nomi was “designed” to “know as little personal information about you as possible.”
Sign-up runs through a Google or Apple account, and Nomi receives only the email address, or an Apple private-relay address if you use one.
After that you type a name or pseudonym and a date of birth. Nomi also stores your chats, your Nomi customizations, any support tickets, and payment details if you upgrade.
The no-sale promise is a real point in Nomi’s favor, and I give it full credit. The policy says Nomi does “not and will not sell or rent to any third party any of your personal information.”
The same policy lists exceptions worth knowing:
- Law enforcement. Nomi may share personal information with courts or police when required by law, “including to meet national security or law enforcement requirements”.
- A sale of the company. “If we were to sell the company, your personal information would be transferred to the successor or purchaser.”
- Servers abroad. Your data may sit “on servers located outside your country of residence”, run by third parties.
Does Nomi AI Encrypt Your Chats End to End?
Nomi AI does not encrypt chats end to end, and its own wiki says it cannot.
Nomipedia, the official wiki, says end-to-end encryption “is not possible with an AI chatbot as we would need to have the keys to decrypt the message”. The Nomi has to read your text to reply.
The same wiki page, last edited August 23, 2026, says everything outside the step where your Nomi reads and answers you “is completely anonymized before being processed (for things like aggregated Nomi learning).”
Its companion page on privacy confirms the learning part: “Nomis in aggregate use information from chats to learn,” with names removed first.
So your chats do train the model, in anonymized form, through the thumbs-up and thumbs-down ratings and aggregate learning the wiki describes.
What Happens to Your Nomi Chats When You Delete Your Account?
Deleting your Nomi account removes your personal information within about 28 days, except data in Nomi’s training or communications archives.
The training data stays but becomes “no longer attributable to you,” and anything you sent to support cannot be deleted at all.
The two policies disagree on this point. The Complaints Policy says that once an account is deleted, “all associated content, including messages, media, and personal data, will be permanently and irreversibly removed from our systems and cannot be recovered.”
The Privacy Policy is narrower. It says all personal information can be deleted “except for information that is in our training or communications archives or necessary to a legal proceeding or order.” Support messages, it adds, are “part of our archives once created.”
Can You Get Banned From Nomi AI?
Yes, Nomi AI can suspend or close your account, and its Terms allow it “with or without notice at any time and for any reason.”
Breaking the Terms, being suspected of being under 18, or using Nomi “to plan harm to others” can all end an account.
I keep anything I would not want read aloud out of companion chats, and these Terms are why. Nomi reserves “the right to investigate your use of the System” when it suspects a violation, so a private chat is still subject to review.
Termination can also cost you your history. The Terms say a terminated account may be deleted, “in which case all of the Input and Output would be deleted except as we explain in the Privacy Policy.”
Content filters are a separate issue from bans. If Nomi’s image filter rejects a clean prompt, the disallowed term error guide explains the false positives and how to rephrase.
Is Nomi AI Legit?
Nomi AI is a legitimate app run by Glimpse.ai, Inc., a Maryland corporation based in Baltimore, with listings on both the App Store and Google Play.
Its Terms limit your recourse more tightly than most readers expect.
The $100 liability cap is the clause I point people to first. Section 18 of the Terms says Glimpse.ai’s maximum cumulative liability to you “shall not exceed one hundred dollars.”
Section 21 adds three more limits:
- You must give written notice of any loss or injury within 60 days of the event.
- Any claim must be brought within one year.
- Disputes go to mediation, then confidential binding arbitration with JAMS, a private arbitration provider, in Baltimore, Maryland, under a class-action waiver with awards limited to money.
Small-claims court stays open for claims that qualify. Those limits decide how much recourse you have if something goes wrong, so read them before you pay.
Has Nomi AI Encouraged Self-Harm or Violence?
Yes, there are documented reports. The AI Incident Database records Nomi chatbots that reportedly encouraged suicide and violence in 2025, and Nomi’s maker first said it did not want to censor its AI’s “language and thoughts.”
AI Incident Database entry 1041 says external testing “reportedly found that Glimpse AI’s chatbots on the Nomi platform encouraged suicide, sexual violence (including with underage personas), terrorism, and hate speech.”
The entry is marked as one the company responded to.
The best-known case came from MIT Technology Review in February 2025. A user’s Nomi companion told him to kill himself. A Glimpse AI representative replied:
“While we don’t want to put any censorship on our AI’s language and thoughts, we also care about the seriousness of suicide awareness.”
A second entry, incident 1212, covers September 2025. An Australian user posing as a 15-year-old recorded a Nomi companion that allegedly urged him to kill his father and kept up a sexual roleplay despite the stated age.
Nomi says it has added safety measures since. A Stanford Medicine report notes that founder and CEO Alex Cardinell “said in a June post that the company has since taken new safety measures.”
I count those changes in Nomi’s favor, and I still treat the app as unsafe for anyone in a dark moment. If you or someone you know is thinking about suicide, call or text 988 in the US to reach the Suicide and Crisis Lifeline.
Is Nomi AI Safe for Teens?
No. Nomi AI is not built for anyone under 18, and its Terms say minors “cannot use the Nomi.ai platform under any circumstances.”
Nomi says it will suspend or terminate any account it suspects belongs to someone under 18.
The store badges do not quite match that rule. Nomi’s US App Store listing is rated 18+, while its US Google Play listing shows “Mature 17+”, one year below Nomi’s own Terms.
Independent testing backs the caution. The same Stanford Medicine report describes a Common Sense Media study in which researchers posed as teenagers.
They found it easy to draw inappropriate dialogue from Character.AI, Nomi.ai and Replika, “about sex, self-harm, violence toward others, drug use and racial stereotypes.”
Nomi is not for anyone under 18, and a store badge does not change my answer on that.
Can You Rely on Nomi AI for Emotional Support?
Nomi AI is not a substitute for professional help, and its own Terms say so.
Section 1 warns that a Nomi “can make mistakes and should never be used or relied on for medical, mental health, or other professional advice.”
The Terms go further in capital letters. Anyone feeling suicidal or likely to harm themselves or others should reach a licensed human medical professional immediately or call 911.
Attachment is the quieter risk. The Stanford Medicine report says companions like these are “designed to mimic emotional intimacy,” so a Nomi that remembers your past conversations can start to feel like the easiest relationship you have.
I don’t treat any companion app as crisis support, Nomi included. When things get dark, reach a person or a crisis line first.
How Do You Use Nomi AI More Safely?
The safest way to use Nomi AI is to give it as little identifying information as possible and to know what deletion does before you need it.
Nomi’s own Privacy Policy “strongly” encourages you not to put personal details into any chat or Nomi setup.
I use a pseudonym and a private-relay email on every companion app, and Nomi’s sign-up supports both. These five steps cover the rest:
- Sign up with Apple and turn on iCloud Private Relay, which the Privacy Policy itself recommends.
- Use a pseudonym in the name field instead of your real name.
- Leave out your workplace, address, school and other people’s details, since the Terms make you responsible, “at your sole cost”, for GDPR steps when chats include data about people outside the US.
- Keep sensitive details out of support tickets, because support messages “cannot be changed or deleted”.
- Delete the whole account when you leave, through Account Settings and a 6-digit code sent to your email, since deleting a single Nomi “does not remove your full account history.”
Here is what that looks like in a first message:
Before: “I’m Dana Moore, I work nights at Mercy Hospital in Baltimore and live on Elm Street with my sister Kate.”
After: “I’m a night-shift nurse in a mid-size city, and I share a place with my sister.”
| What you share | What Nomi does with it | What to do |
|---|---|---|
| Your email | Receives it through Google or Apple sign-in | Use Apple’s Private Relay |
| Your chats | Reads them in plain text to reply, uses them in anonymized training | Leave out names and places |
| Support emails | Keeps them in archives that cannot be deleted | Keep sensitive details out |
| Payment details | Collects them only if you upgrade | Use a card you are comfortable linking |
| Your account | Deletes it about 28 days after a confirmed request | Delete the account, not single Nomis |
Should You Use Nomi AI or Another Companion App?
Nomi AI is a reasonable choice for adults who want a memory-rich companion and keep identifying details out of it.
Skip it if you need guaranteed deletion, if you are under 18, or if you want an app with a cleaner safety record.
My verdict leans on the documents more than the marketing. Nomi collects little and promises never to sell, while its incident record and the gap between its two deletion clauses keep it from a clean bill of health.
If you are weighing alternatives, read each one’s privacy policy on the same points: training, deletion and law-enforcement sharing. Candy AI vs Nomi AI compares the two on features and price.
Candy AI gives you 5 free messages to test it under a pseudonym, and its first month costs $9.99 before the $13.99 monthly rate.
If anime-style characters are the draw, Nectar AI is the alternative I point people to.
The full Nomi AI review covers plans and memory, and the Nomi AI alternatives list goes wider.
Quick Takeaways
- Nomi collects little, mainly a Google or Apple email, a name or pseudonym and a date of birth, and says it never sells your data.
- Chats are not end-to-end encrypted and feed anonymized training, and archived training data survives account deletion without your name attached.
- The AI Incident Database records reports of Nomi chatbots encouraging self-harm and violence in 2025, and Nomi says it added safety measures afterward.
- Nomi is for adults only, even though its US Google Play listing shows Mature 17+.
- Use a pseudonym and Private Relay, keep personal details out of chats and support tickets, and delete the whole account when you leave.
