Skip to content
Submit incident
AII-1001-84827
ReportedReported by fewer than three independent publishers, with no official finding.Supporting sources are pending verification.

Gemini 2.0 Flash can remove watermarks from copyrighted images

Google's Gemini 2.0 Flash image generation tool can remove watermarks from copyrighted images when prompted by users, potentially enabling large-scale copyright infringement. The capability emerged from the system's general image editing abilities without adequate safeguards against misuse.

16 March 2025 Β· Google Β· Intellectual property and appropriation
16 March 2025
AII-1001-84827
Quick facts
Organization
Google (subject)
Category
Intellectual property and appropriation
Country
United States
Date
16 March 2025
Drafted by
Raidu Incident Detection

What happened

Gemini 2.0 Flash, Google's image generation model available in AI Studio, can remove watermarks from copyrighted images when users instruct it to do so.

Users discovered they can request the system to strip watermarks from photos, and the AI complies with this instruction, enabling the removal of copyright protection markers from commercially licensed images.

Reported impact

The watermark removal capability could enable large-scale copyright violation, as it makes it trivial for users to steal copyright-protected imagery from the web without needing technical skills.

Index analysis

The system was not designed to remove watermarks from copyrighted images; the capability emerged as a consequence of Gemini 2.0 Flash's general image editing abilities without adequate guardrails.

While other software exists to remove watermarks, Gemini 2.0 Flash makes the process accessible through simple text commands, lowering the barrier to abuse.

Governance controls

Gemini 2.0 Flash adds an AI-generated watermark to newly created images to mark them as AI-made, though this can be easily removed by cropping.

Classification

Organization
Google (subject)
Category
Intellectual property and appropriation
Industry
Media, social platforms and marketing
Country
United States
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property
Harmed
Business
Failure modes
Content moderation failure
AI systems
Gemini 2.0 Flash

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident