DJI Romo Cloud authorization bug exposes 7,000 robot vacuums
A backend permission validation error in DJI's cloud servers allowed unauthorized access to thousands of DJI Romo robot vacuums. The vulnerability exposed live camera feeds, microphone audio, and home maps to any authenticated user.
Once you’re an authenticated client on the MQTT broker, if there are no proper topic-level access controls (ACLs), you can subscribe to wildcard topics and see all messages from all devices in plaintext at the application layer.
Key facts
- What
- A backend permission validation error in DJI's cloud servers allowed unauthorized access to thousands of DJI Romo robot vacuums.
- Incident date
- Feb 8, 2026
- Who
- DJI
- Failure mode
- Data Leakage
- AI surface
- Autonomous System
- Severity
- High
What happened
A researcher discovered that a single private token could access data from nearly 7,000 DJI Romo vacuums across 24 countries. The exposure included live video, audio, and detailed 2D floor plans of users' homes. DJI deployed an automatic patch in early February 2026 to resolve the issue.
What broke inside the model
- 01 · TriggerA request triggers retrieval or context loading.
- 02 · Model stepThe context pulls in another user's content.
- 03 · Control gapNo boundary enforces isolation at the moment of output.
- 04 · FailurePrivate data crosses into the response.
- 05 · ConsequenceOne user sees another's data, and disclosure follows.
One user's content crosses the retrieval boundary into another's response.
The system failed to implement proper topic-level access controls (ACLs) on its MQTT brokers. This allowed authenticated clients to use wildcard subscriptions to receive data packets from any device on the network.
What it cost
Sources
Cite this entry
https://failureindex.ai/failures/dji-romo-cloud-authorization-bug-exposesAI Failure Index. "DJI Romo Cloud authorization bug exposes 7,000 robot vacuums" (FI-0555). Realm Labs. https://failureindex.ai/failures/dji-romo-cloud-authorization-bug-exposes (indexed Jun 16, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0555. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.