Cases › security breach › LR-2026-0412

Man accidentally gains access to thousands of robot vacuums

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

Status: action reported, no fix

Severity moderate

Reported 25 Feb 2026

No source states: loss or records, developer and model, agent tool, could do, deployer, industry.

Severity moderate: Test: data or systems of others touched with a bounded scope stated. Fact: nearly 7,000 robot vacuums across 24 countries, including live camera feeds and microphones of other people's homes

A software engineer, Sammy Azdoufal, was steering a DJI Romo robot vacuum with a game controller when he used an AI coding assistant to reverse-engineer how the vacuum talked to DJI's remote servers. He extracted a security token meant to prove ownership of his own device. The backend servers instead treated him as the owner of nearly 7,000 robot vacuums across 24 countries, and he could reach live camera feeds, microphones and floor plans of other people's homes. He responsibly reported the security bug to The Verge rather than exploiting it. Press coverage used the case to argue that automated systems are widening security gaps. The sources do not state how DJI fixed the flaw or when it was closed.

What happened

1 · What went wrongHe extracted a security token meant to prove ownership of his own device.

2 · What it reached or didThe backend servers treated him as the owner of nearly 7,000 robot vacuums across 24 countries.

What was done

He responsibly reported the security bug to The Verge rather than exploiting it.

Risks

Credential and privilege abuse
Primary37 incidents

Privacy harm to a person
Also41 incidents

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

Controls

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

Application

Short-lived, scoped credentials · Secrets kept out of the model's context · Separate identity for each agent · Least-privilege access for the agent · Alerts on unusual tool use · Activity logs · Sensitive-data checks on outbound data

Infrastructure / target

Multi-factor sign-in and password policy · Secrets scanning

Evidence

No source states: what the AI was doing, how it was caught, when it happened, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off.

Show the full account: 1 items from the primary sources

What happened, in order

Not dated in the sources:

He reported the security bug to The Verge rather than exploiting it.

Show every field, with its source and grade