tech
Exclusive: Grok also falls for jailbreak that tricked ChatGPT into creating sexual, graphic images
The prompt that Axios reviewed suggests that relatively simple jailbreaks can still bypass AI image safeguards.
TL;DR
- A jailbreak prompt that tricked ChatGPT into generating graphic images also works on SpaceXAI's Grok.
- Researchers found Grok could produce nude women or bloody body parts with minimal prompting.
- The same technique could be used to create deepfakes, significantly reducing the effort required for fake image generation.
- SpaceXAI's terms of service mention that outputs could be sexual or violent based on user input.
- While Grok's outputs were more stylized than ChatGPT's, the jailbreak remained effective.
- The findings add to scrutiny over Grok's image safeguards amid an ongoing lawsuit.