On Friday, several media reports reported that X had restricted the editing of Grok AI pictures on the platform, with some users being told that it could be opened by paying $8. This was seen as a response to the recent dispute over Grok, an AI robot that created a proliferation of pornographic images of women and children on the platform.
“Image generation and editing functions are currently limited to fee-paying subscribers,” is described in Grok’s reminder to users. However, according to the media The Verge, unsubscribed users can still edit images using Grok on the desktop website. The X platform appears to limit only the way in which users can request editing through Grok, but a mobile end user can still trigger editing by pressing any picture. Hide Paths and Security GapsThe non-publicly requested image-editing content does not enter the public information stream, which means that X is anxious to resolve only to prevent Grok from directly posting harmful images on the platform. The company has not responded whether or not it is repairing these loopholes, but there are precedents for imperfect updates since Elon Mask acquired Twitter. In addition, well-motivated users can still produce abusive content free of charge through independent Group applications or websites.
According to previous BBC reports, some of the outlaws have promoted child sexual abuse materials allegedly generated by Grok on the net. Connected magazine this week revealed that the Grok application and the images generated by website users are more visible and disturbing than those created by X platform users. Can the pay wall cure it?It is not clear how a charge-back mechanism could prevent the output of controversial content, as the security code in Grok is still in question. The chat robot is still set to acquiesce to “good faith” when a user requests a “youth” female image, and XAI argues that this expression “does not necessarily mean a minor”. This may continue to result in Grok producing harmful images of minors. Other policy priorities of xAI include requiring Grok not to preach ethically to users and not to limit the “fictional content of humanity that relates to the subject of darkness and violence”. An AI security expert indicated to Ars Technica that Grok could have been adapted to a safer model, and that its existing security guidelines were like the platform’s design of both. Global regulatory pressure escalationThis update comes at a time when X is at risk of fines and legal action by global regulators. X seems to want to reduce the generation of illegal content by forcing users to provide identity information and credit card data, and when advocates against image sexual exploitation point out that, even if not in some states, Grok’s “dressing” function can cause lasting psychological, financial and reputational damage to victims. This means that paid users may continue to use Grok for the production of harmful content, while X may not be allowed to do so because it is not illegal at the technical level. In 2024, X undertook to review all involuntary private images on a voluntary basis, but the promotion of Mask images of swimsuits by the public and private figures indicated that the policy was no longer valid.
The U.S. regulator is currently silent on the illegal export of Grok, but Friday the Democrat Senator began pressuring Apple and Google to lower X and Grok to enhance their security. In a letter to Apple CEO Couk and Google CEO Sandhar Pichai, the Senator wrote: “X’s knowledge of these trends and, at least, his negligent response cannot be ignored. Condoning X will be a mockery of your audit practice.” The British Prime Minister, Kiir Stammer, went so far as to say that the bad output of Grok “is illegal and we will not tolerate”. He has asked that “all viable options be put on the table” and warned that “X must correct and remove those elements, and we will act”. In an interview with The Guardian, British Congressman Jess Asato stated that even the introduction of a pay-for-wall would not be sufficient to put an end to the censorship: “The addition of pornography, bullet holes or bikini to female images is still a digital violation and XAI should permanently disable the function.”
At present, X may have restricted public access only to the scope of Grok ‘ s non-compliance, but has failed to address the problem at all. According to the online magazine, Grok even promotes the mainstreaming of nudity/stripping applications and companies profit from this function. The British Communications Authority may face a British ban or a fine of up to 10 per cent of the global turnover if it finds that the Mask chat robots are violating the Cybersecurity Act.
