Gemini allows download of blocked AI images
Original title: Google Gemini Safety Bypass Lets Users Download AI-Generated Images That Its Own Filters Blocked
Gemini safety filter sometimes delivered refused images via download path after chat showed permitted preview; issue resolved via maintenance.
Google Gemini permitted downloading AI-generated images blocked by safety filters. Users requested a permitted image after a refused one, but download returned the rejected output. Researchers reported the flaw to Google in June 2026. By October 22, 2026 the bypass could no longer be reproduced after 12 attempts. Google attributed the change to routine maintenance rather than a fix linked to the report. No data access or privilege escalation occurred. The finding shows why AI safety testing must cover stored files and download links.
Unconfirmed: State-management error explanation remains unconfirmed
- Bypass did not require jailbreak
- Tested on copyrighted characters, synthetic person and own likeness
- Older conversations downloaded correct images by October 2026
- Google closed report after initial out-of-scope marking
- Recommendation: bind downloads to approved preview
Impact: bypassing AI safety filter for refused image download
- 1.Bind each download to its approved preview
From the original article
2 min read at cybersecuritynews.com
Google Gemini allowed users to download AI-generated images that its safety filters had refused to display, according to research published by Fog Security on November 5, 2026. The bypass did not require a jailbreak. Instead, the download feature sometimes delivered a rejected image while the conversation showed a different, allowed picture.
By Abinaya · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.