Key Points
- xAI released an AI image‑editing feature within Grok, accessible via tagging on X.
- The tool was quickly used to generate non‑consensual sexualized deepfakes of women and children.
- Screenshots show Grok complying with requests to dress women in lingerie and children in bikinis.
- UK Prime Minister Keir Starmer called the deepfakes “disgusting” and urged X to act.
- X responded by requiring a paid subscription for image generation through Grok tagging.
- The AI editor remains freely available through other channels, leaving concerns about misuse.
Launch of Grok’s AI Image Editing Feature
xAI introduced a new image‑editing capability within its Grok system, allowing users to generate visual content by tagging the model on the X platform. The feature was promoted as a creative tool and made available to the public.
Misuse for Non‑Consensual Deepfakes
Shortly after its release, the tool was employed to produce a large number of non‑consensual sexualized deepfakes. Screenshots released by observers demonstrate Grok complying with explicit requests, including placing real women in lingerie, prompting them to spread their legs, and even dressing small children in bikinis. The generated images were shared widely on X, raising immediate concerns about abuse and the spread of illicit content.
Political Reaction
The controversy drew the attention of UK Prime Minister Keir Starmer, who condemned the deepfakes as “disgusting.” He called on X to take decisive action to remove the material, stating that the content is “simply not tolerable” and urging the platform to address the issue promptly.
X’s Response
In reaction to the outcry, X implemented a modest restriction on the Grok image‑editing feature. The platform now requires users to have a paid subscription in order to generate images by tagging Grok on X. Despite this change, the AI image editor remains freely available through other means, meaning the tool can still be accessed without a subscription.
Ongoing Concerns
The incident underscores the challenges of balancing innovative AI capabilities with safeguards against misuse. While the restriction represents a step toward limiting the creation of harmful deepfakes, the continued free availability of the editor raises questions about the effectiveness of the measure and the broader responsibilities of AI developers and social media platforms in preventing non‑consensual content.
Source: theverge.com