Technology · Adult AI Companion · Consumer

My Lovely AI Data Breach (2026)

NSFW 'AI girlfriend' companion platform.

Pilot record · BRI 2.0 results pending
Breach Risk Index
LowGuardedElevatedHighSevere
Not yet scored
Circulation not yet audited
What this rating measures

The Index rates how dangerous a breach is right now rather than how large it was. It values each exposed field on its own, ages each one according to what that field is, and models what the assembled set lets an attacker do.

It is read against what is observed circulating, not what was announced. Five bands run from Low through Guarded, Elevated and High to Severe.

Every rating carries a status. Observed means it rests on fields seen circulating. Observed minimum means at least that much was seen, so read it as a floor. Provisional means the field list comes from breach catalogues and is not yet verified. How sure we are of the evidence is shown separately, and it never lowers the rating. How the Index works →

Data sensitivity
Restricted

Includes data whose disclosure is itself the harm, such as sexual orientation or intimate preferences. It cannot be taken back, and it is used for extortion and outing.

Volume
Count not yet verified
Field types
6 listed
Breach date
Apr 2026
Vector
Database Exposure
How it unfoldedListed the same month as the breach
  1. Breach
  2. Listed in breach catalogues
What came out

What data was exposed

Sexual Content (prompts)Cannot be reissued
Email AddressContact
Media URLContent
Social Handle (discord)Social
Social Handle (x)Social
User IDCredential
Red cannot be reissued or taken back · Blue is the residence · Black is movement, where you go rather than where you live · Grey is high-signal data that is not standard breach material · Field list from breach catalogues, not yet verified
What it enables

How this data can be used against you

Assessed once the record is scored.

What you should do

One key action to take

Treat messages about My Lovely AI with care

Do not act on a call, email or text about this breach. Contact My Lovely AI using a number you already have.

Check it yourself

Check whether you were affected

Most people in a breach like this do not need an engagement. They need to know whether their address is in it, and when it turns up somewhere new. HackNotice does both, free: an Instant Check now, alerts after.

HackNoticeRuns on HackNotice · ObscureIQ never sees the address you check
Run the Instant Check →
If this matters to you

Talk to us about your exposure

If you are exposed in a way that cannot be rotated away, or you are responsible for someone who is, a call establishes what is actually reachable about you and what is worth doing first. No charge for the conversation.

Request a confidential call →
Full record · 9 sections

The full breach analysis

What happened, what the company reported against what is circulating, the field-by-field record, the actor and the advisory.

  • What happened
  • Recent developments
  • Reported and circulating
  • Data points exposed
  • Downstream risk
  • Principal risk advisory
  • Threat actor
  • Background
  • Data context
No sign-in · nothing is recorded
What happened

How this breach happened

In April 2026 My Lovely AI was breached, exposing over 100,000 users. Two datasets totalling about 113,000 records held explicit NSFW prompt text, with nearly 70,000 messages linked to unique user IDs, plus links to generated images and a small number of Discord and X usernames.

Recent developments

What has happened since the breach

The platform's data was exposed in April 2026, surfacing explicit user content tied to identities.

Reported and circulating

What was reported vs. what is circulating

The company's notice

Reported

Status
Not yet researched
Not yet verified by ObscureIQ

Listed in breach catalogues

Breach date
Apr 2026
Volume
Count not yet verified
Fields listed
6email address, user ID, sexual content (prompts), media URL, social handle (discord), social handle (x)

The score is built only on what is listed in circulation.

Data points exposed

Every exposed field, and how long it stays useful

FieldShelf lifeIn noticeListed
Email AddressLasts for yearsNoYes
User IDLasts for yearsNoYes
Sexual Content (prompts)Lasts for yearsNoYes
Media URLLasts for yearsNoYes
Social Handle (discord)Lasts for yearsNoYes
Social Handle (x)Lasts for yearsNoYes

Shelf life is how long a field stays useful to someone holding it. A password expires fast. A government ID number or a date of birth lasts for years. An address stays useful long after it stops being current.

Downstream risk

How attackers can use this data

Explicit prompts tied to user IDs and social handles are highly coercible; the primary risk is sextortion, blackmail and doxxing rather than ordinary identity fraud.

Principal risk advisory

Risks for executives and other high-profile people

An intimate-data breach: preferences, orientation or explicit content linked to an identity create acute coercion and blackmail exposure. For a high-profile principal this is targeting-grade, not merely identity-theft-grade: the combination lets an adversary locate, impersonate, or pressure the principal with little additional work.

Threat actor

Who is behind this breach

No threat actor has been publicly attributed.

Background

About the company

My Lovely AI is an NSFW AI-companion ('AI girlfriend') platform that generates adult text and imagery from user prompts.

Data context

Why the company holds this data

An AI-companion platform holds user account and contact data, user-generated prompts, generated media, and linked social handles.

Database ExposureTechnologyAI ServicesPrivate company2026

About this record

No reported total is on record for this breach. No count has been verified for this breach yet. Figures in older breach catalogues are not shown until they are checked against the data. No score is shown until the circulating field list has been audited. Scores come from the Breach Risk Index, which rates what the exposed fields still let an attacker do, not how large the breach was. Breach records change as evidence does. If you hold a notice, a filing or a dataset observation this page does not reflect, report a correction.

ObscureIQ · Circulating breach record How the Index works · Contact MMXXVI