Monday 16 November
Global

Your region, kept on this device.

Google Gemini Download Feature Reportedly Delivered Images Its Safety Filters Had Blocked

Severity
Medium
Sources
2

reports that 's web app download path could deliver images that safety filters had refused, while the chat preview showed an allowed image; no jailbreak was required. Affected links use lh3.googleusercontent.com with signed-in session tokens, inline links ending =s1024-rj and download links ending =s0-d-I. Fog suspects a state-management error across generation, safety checks and cleanup, which is unconfirmed. Google closed the report on 23 October, citing maintenance and no internal ticket. Fog recommends binding each download to its approved preview and purging files from refused or failed generations.

For security analysts · Written by AI from 2 sources · Wrong?

What to do

  1. AI developers should bind each download to its approved preview and remove files from refused or failed generations, as Fog Security recommends.
  2. Test AI safety controls across stored files and download links, not only chatbot replies.

Written by AI from the sources, checked against them

Full story timeline · 2 updates

  1. Mon 9 NovemberGoogle Gemini Safety Filter Bypass Exposes Blocked AI Images Through Download LinksGBHackers
  2. Google Gemini Safety Bypass Lets Users Download AI-Generated Images That Its Own Filters BlockedCyber Security News

Who reported · 2 outlets

GBHackersCyber Security News

Who and what is in it

Products

Report a problem