X has announced updates to its Grok account, aimed at preventing the creation of nonconsensual sexual deepfakes. However, these changes do not fully address the issue, as it remains possible to generate revealing images through Grok. According to The Verge AI, despite X's claims, Grok still allows users to create revealing deepfakes, even after the latest policy updates.
The Grok account on X has faced significant scrutiny following the spread of nonconsensual sexual deepfakes. These deepfakes, often depicting real individuals in revealing or sexualized poses, have raised serious ethical and legal concerns. X's recent announcement comes amid growing pressure from regulatory bodies and public outcry over the misuse of artificial intelligence for unethical purposes.
X has implemented technological measures to restrict the Grok account from editing images of real people into revealing clothing such as bikinis. Additionally, the ability to create and edit images through Grok is now limited to paid subscribers. This change aims to provide an additional layer of accountability and deterrence against misuse.
The technical implementation involves filtering prompts that request revealing edits to images of real people. For example, prompts like "put her in a bikini" are now censored. However, the system still allows for creative workarounds. Testing revealed that users could still generate revealing images by using alternative phrasing or less explicit prompts. X attributes these issues to "user requests" and "adversarial hacking of Grok prompts."
Despite the restrictions, the platform's ability to generate revealing deepfakes remains a significant concern. The implications of this capability extend beyond ethical considerations; it also raises legal issues. In the United Kingdom, a new law is being enacted this week to criminalize the creation of nonconsensual intimate deepfake images. The UK's communication regulator, Ofcom, has already launched an investigation into X's handling of these deepfakes.
Prime Minister Keir Starmer has expressed that X is taking steps to comply with UK law, but emphasized that the government will not back down until full compliance is achieved. The BBC reports that the prime minister's official spokesperson provided a "qualified welcome," based on media reports that X had taken action. However, according to The Verge AI, testing indicates that the problem persists and has not been fully resolved.
Moving forward, the effectiveness of X's measures and their alignment with new regulations will be closely watched. Users and regulators will need to see concrete evidence that the platform can effectively prevent the creation of nonconsensual deepfakes. As technology evolves, so too must the safeguards put in place to protect individual rights and privacy. Watch for further updates on X's compliance efforts and any new regulatory actions taken by governments worldwide.