No, Grok hasn’t paywalled its deepfake image feature

Elon Musk’s X has partially restricted entree to Grok’s image editing capabilities amid increasing backlash to the flood of nonconsensual, sexualized deepfakes of adults and minors generated by the platform. As of this writing, it nary longer generates images arsenic @grok replies for free, but Grok’s image editing devices stay readily disposable for immoderate X personification to churn retired images, some sexualized and tame.

X users were antecedently capable to inquire Grok – by tagging @grok successful a tweet – to edit aliases create images connected the level by tagging it successful a post. Users now attempting this are met pinch an automated consequence from the chatbot’s account, telling them that “Image procreation and editing are presently constricted to paying subscribers.” The reply includes a nexus encouraging them to subscribe to X’s paid programs “to unlock these features.”

The connection gives the belief that only paid X users tin edit aliases create images utilizing Grok. That connection is reinforced successful the galore headlines circulating correct now stating that xAI, different Musk institution that makes Grok, has restricted Grok’s image editing abilities to this prime group of users. But that belief is false. All X users — including free ones — tin still usage Grok to edit and create images, including the sexually suggestive deepfakes that landed Grok successful basking h2o to statesman with.

Interacting pinch Grok done replies connected X is conscionable 1 of respective ways to usage the AI chatbot. For example, the “Edit image” fastener still invokes Grok connected each image displayed connected X’s desktop website, and that aforesaid fastener tin beryllium accessed connected the X app by long-pressing connected immoderate image. Grok tin besides beryllium accessed done a standalone website aliases app, arsenic good arsenic via a highly promoted tab successful the X apps and X website. These stay readily disposable to free users, and require nary further effort compared to the @grok command.

The Verge tested the Grok website, app, and X tab pinch free accounts and Grok readily complied pinch our requests to edit aliases make a assortment of images, including a afloat “nudify” and 1 of Musk wearing a bikini.

The restrictions connected X travel aft Grok has been utilized to make intersexual images chiefly of women, galore of them real, arsenic good arsenic minors. Much of this landed connected X, which the Financial Times dubbed “the deepfake porn tract formerly known arsenic Twitter.” The deepfake undressings, arsenic good arsenic different intersexual material, person outraged regulators worldwide, who person threatened to return action against X, though the deepfakes whitethorn autumn into something of a ineligible void.

Musk and xAI person besides threatened to return action against users who create forbidden contented utilizing Grok, alternatively than ensuring that guardrails beryllium to make it intolerable to create it successful the first place.

X’s attack — limiting entree alternatively than constraining usage — contrasts pinch the attack taken by different AI companies for illustration Google and OpenAI, which thin to enforce strict guardrails connected devices for illustration Gemini and ChatGPT, alternatively than let free rein for those who pay. Musk has reportedly opposed stricter guardrails personally, and respective members of xAI’s already sparse information squad reportedly discontinue successful the lead up to the deluge of integer undressings.

X did not respond to The Verge’s petition for comment.