Sexual deepfakes: First state lifts ban on Grok access

The AI company xAI has convinced the responsible authority in the Philippines that the AI chatbot Grok can be sufficiently modified to prevent sexualized deepfakes. In return, a ban on the AI software imposed a week ago was lifted, reports the financial news agency Bloomberg, citing the responsible cybercrime center. xAI has pledged to modify the tool in this regard, which is why access to it has been reinstated. Last week, the Philippines became the third country in Southeast Asia to block access to Grok. Indonesia and Malaysia have not yet reversed this decision.

Different Reactions

The move by the authority in Manila shows that xAI can succeed in defusing the sometimes massive criticism of Grok. However, this does not seem to be working everywhere, as has become apparent this week. On Monday, the Handelsblatt reported that the EU Commission intends to take measures against Grok. If it fails to completely prevent the generation of unauthorized sexualized deepfakes of individuals, the service could be banned in Europe, the newspaper quoted. xAI had already announced at the time that a technological block should prevent such AI images. However, this did not work consistently at first.

The background to the state reactions are AI images that Grok generated, partly in the thousands per hour, at the beginning of the year. These primarily depicted real women and sometimes minors in underwear or bikinis. After the initial strong criticism, xAI restricted the technology to the extent that image generation was only possible in direct chats or for a fee. However, this was widely interpreted as the problematic function simply being marketed as a premium offer. Therefore, xAI made another move last week and promised to completely stop this type of image generation.

(mho)