Briefly
- The EU has launched a probe into whether or not X correctly assessed dangers earlier than deploying Grok’s AI options.
- The investigation will study X’s compliance with the area’s Digital Companies Act obligations.
- It marks Europe’s newest crackdown on AI-generated deepfakes, with a number of international locations banning Grok over youngster security considerations.
The European Fee launched a proper investigation Monday into whether or not X violated EU digital guidelines by allegedly failing to stop its Grok AI chatbot from producing and spreading unlawful content material, together with sexually specific photographs of kids.
The probe will assess whether or not the corporate correctly evaluated and mitigated dangers earlier than deploying Grok’s picture era options, the Fee mentioned in a Monday assertion.
The Fee additionally mentioned dangers have materialized by way of the precise era and unfold of unlawful sexual content material, exposing EU residents to severe hurt.
It comes amid mounting worldwide scrutiny of Grok’s function in creating non-consensual deepfakes.
Two weeks in the past, X applied restrictions, limiting picture era to paid subscribers, and added technical obstacles to stop customers from digitally manipulating folks into revealing clothes. The corporate additionally geoblocked the function in jurisdictions the place such content material is against the law.
Regardless of these measures, researchers discovered that about one-third of the sexualized photographs of kids recognized within the CCDH pattern remained accessible on X’s platform.
“With this investigation, we’ll decide whether or not X has met its authorized obligations underneath the DSA, or whether or not it handled rights of European residents—together with these of girls and kids—as collateral harm of its service,” Henna Virkkunen, Government Vice-President for Tech Sovereignty, Safety and Democracy, mentioned within the assertion.
Earlier this month, EU Fee spokesperson Thomas Regnier condemned X’s “Spicy Mode” function at a Brussels press convention.
“This isn’t spicy. That is unlawful. That is appalling. That is disgusting. This has no place in Europe,” Regnier mentioned.
Fraser Edwards, co-founder and CEO of cheqd, advised Decrypt that “each creator ought to be capable to management how their likeness is utilized in AI-generated media.”
He says the “backlash round deepfake abuse underscores a fundamental failure of the web itself.”
“There may be nonetheless no native option to confirm who created a bit of artificial content material or whether or not its use was ever authorised”, Edwards added, leaving legal responsibility to proceed “defaulting to intermediaries like X reasonably than the folks liable for producing the abuse.”
If confirmed, the failures underneath investigation would represent infringements of Articles 34(1) and (2), 35(1), and 42(2) of the Digital Companies Act, which require platforms to evaluate and mitigate systemic dangers, together with unlawful content material dissemination and unfavourable results associated to gender-based violence.
The EU investigation extends a late-2023 DSA case that resulted in a $140 million (€120 million) fantastic towards X in December for misleading design, advert transparency failures, and restricted researcher entry.
The Fee has since expanded scrutiny to Grok, together with prior considerations over antisemitic content material generated by the chatbot.
Decrypt has reached out to xAI for additional remark.
Each day Debrief E-newsletter
Begin day-after-day with the highest information tales proper now, plus authentic options, a podcast, movies and extra.