RISE Jailbreak Attack Bypasses Safety Filters of OpenAI and Google Image Models
Understand the story
On October 14, 2026, Ars Technica reported that the security research firm White Circle proposed a jailbreak attack method called RISE in a preprint paper. The method uses an off-the-shelf downloadable Qwen model as the attacker and Gemini 3.8 Flash as the judge, making API calls to bypass the safety filters of OpenAI's DALL·E 3 and GPT-Image-2 as well as Google's Nano Banana 2 image generation models. According to the study, about 13% of attack requests successfully made these models generate explicit content. The results show that even the latest commercial image generation models may have their safety mechanisms against inappropriate content breached by automated attack methods. No responses or fixes from OpenAI, Google, or other relevant companies have been reported yet.
AI overview of reporting · Updated 4 days ago
Reporting timeline
Explore different aspects of the story through its reporting.
- Ars Technica · AIOpenAI, Google pushing new AI models easily tricked into making nudes, researchers say
Security research firm White Circle has proposed an attack method called RISE in a preprint paper, using an off-the-shelf downloadable Qwen model as attacker and Gemini 3.8 Flash as judge, bypassing the safety filters of DALL·E 3, Google Nano Banana 2 and GPT-Image-2 via API calls, with about 13% of requests successfully generating explicit content.
Story heat trend
Current heat3·Comparable peak10（10/14 20:00）·Comparable change over 24 hours -50%
The trend compares the same continuously observed sources, which may cover fewer sources than the current heat score. Hover or select the chart to inspect each hour; use the left and right arrow keys to navigate.