AI Risk Wire

Cases › leaked private data › LR-2026-0363

Exposed database of AI image generator apps leaked over one million images

Preliminary Why: 1 report: WIRED · subject to change · 1 independent report

Status: action reported, no fix

Severity moderate
Loss or records 1,099,985 records
Reported 5 Dec 2025
Developer and model DreamX

No source states: agent tool, could do, deployer, industry.

Severity moderate: Test: data of others touched with a bounded scope stated. Fact: the exposed database contained 1,099,985 records.

An AI image generator startup, DreamX, which operates the MagicEdit and DreamPal apps, left a database exposed and accessible to anyone online. The database held more than one million images and videos made with its tools, most of them pornographic, and some appearing to depict children. Security researcher Jeremiah Fowler found the unsecured database in October and reported it. After he contacted the company, it closed access to the database and launched an internal investigation, and the apps were later removed from app stores. The sources do not say how long the data was left open, and they give no date for the discovery beyond the month.

What happened

1 · What went wrongThe database was left unsecured and accessible to anyone online, a flaw the security researcher Jeremiah Fowler found in October.
2 · What it reached or didSome images appeared fully AI-made, while others looked hyperrealistic and based on real people, according to the researcher.
3 · How it was caughtThe researcher reported the exposed database to the US National Center for Missing and Exploited Children.

What was done

After the researcher contacted the firm, it closed access to the database and launched an internal investigation with external legal counsel.

Risks

Data leaked or sent out
Primary32 incidents
Privacy harm to a person
Also41 incidents
Harmful content and jailbreaks
Also84 incidents
Rising: 6 vs 3.8 a week

Read from the sources; not yet reviewed by a person.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Harness

Network egress control

Application

Sensitive-data checks on outbound data · Output filtering and redaction · Human approval for high-impact actions · Data minimisation · Least-privilege access for the agent · Activity logs

Evidence

No source states: what the AI was doing, when it was found, the direct loss, a count of the other parties affected, any court case, what set it off.

Show the full account: 7 items from the primary sources

What happened, in order

  1. October Security researcher Jeremiah Fowler discovered the unsecured database.
  2. October At the time of discovery, about 10,000 new images were being added to the database every day.

Not dated in the sources:

  • Google suspended MagicEdit and another BoostInsider app for policy violations.
  • Fowler reported the exposed database to the US National Center for Missing and Exploited Children.
  • After Fowler contacted the firm, it closed access to the exposed database and launched an internal investigation.
  • The MagicEdit and DreamPal websites and apps stayed accessible until WIRED contacted those who run them.
  • Apple removed the apps from the App Store.
Show every field, with its source and grade

The AI system

Developer
DreamX
C one secondary source states it
Stage of use
in live use
C one secondary source states it

What happened

Earliest source we hold
2025-12-05
D inferred by us from the described behaviour, reason shown
Who caught it
an independent researcher or journalist
C one secondary source states it
Cause
a shared system was the path
C one secondary source states it
What went wrong
leaked private data
C one secondary source states it
primary risk
RK04
D inferred by us from the described behaviour, reason shown
secondary risks
RK15, RK12
D inferred by us from the described behaviour, reason shown

Impact

Records exposed
1,099,985
C one secondary source states it
Severity
moderate
D inferred by us from the described behaviour, reason shown
Kind of harm
privacy
C one secondary source states it
Who was affected
Innocent people, including underage people, whose images were used without consent in sexual content.
C one secondary source states it
Kinds of data exposed
images, videos
C one secondary source states it
Why this severity
Test: data of others touched with a bounded scope stated. Fact: the exposed database contained 1,099,985 records.
D inferred by us from the described behaviour, reason shown

Notes on gaps

The company's statement that no operational systems were compromised was not used.

LR-2026-0363 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON