Research
AI Guardrail Bypass Research: Comparing ChatGPT and Grok Image Safety
Publication date unknown · Discovered mindgard.ai
Page published
Publication date unknown · First observed: 4 Nov 2026
Coverage timeline
Single-source research — one report is available.
Why it matters
Mindgard’s comparative findings highlight the need to test image-generation safeguards against related jailbreak variants rather than assume a blocked prompt or one model’s behavior generalizes across systems.
Mindgard reports that the same jailbreak prompt and minor variations bypassed image-generation safeguards in ChatGPT, Grok, and other models, producing substantially different harmful outputs. Follow-up testing reportedly found that ChatGPT continued generating objectionable imagery with minor prompt variations.