Following the Wednesday shooting of Slovakia’s prime minister, Robert Fico, European Union enforcers of the bloc’s online governance regime, the Digital Services Act (DSA), said on Thursday that they are closely monitoring disinformation campaigns on the Elon Musk-owned social network X (formerly Twitter).
Along with a host of other issues, the bloc has been officially looking into X since last December for disinformation in civic debate and the efficacy of the platform’s crowdsourced “Community Notes” content moderation function; penalties have not yet been issued.
Musk directly responded to—and thus magnified—a post on X yesterday by right-wing political figure Ian Miles Cheong, who attempted to connect the massacre to the opinions Fico allegedly has, including opposing the World Health Organisation’s pandemic preventive strategy.
When asked to comment on the development during a press Q&A as part of a background briefing the EU held to discuss two Meta DSA probes the EU announced earlier today, a senior Commission official confirmed they are monitoring content on the platform and analysing whether there is “any additional evidence” vis-à-vis the effectiveness of X’s disinformation mitigation measures—to feed the EU’s ongoing investigation.
Reminder: Musk’s shitposting habit may ultimately cost the firm money over regulatory enforcement cycles, since violations of the DSA may result in penalties of up to 6% of worldwide annual turnover.
A Commission source also told Eltrys that Roberto Viola, the director-general of the EU department responsible for regulating communications and technology, had written to X and the other approximately twenty very large online platforms (VLOPs) listed under the DSA, advising caution in light of the attempt on Fico’s life.
The EU wants platforms to be prepared to take corrective action should bad actors try to alter a social media video of the incident in order to take advantage of the circumstances and disseminate false information. Tomorrow’s Commission election roundtable discussion will also require VLOPs to detail the specific measures they are taking to limit the spread or amplification of any modified media related to the incident.
Election Watch Grok
Additionally, on Thursday, X revealed that premium users in the EU may now use their typing fingers on Grok, Musk’s generative AI chatbot, which has been programmed to be fundamentally politically wrong (as opposed to the alleged political correctness of competitors like OpenAI’s ChatGPT). Musk announced the news in a quick post, writing in caveman language, “Grok is now available in Europe.” Going where they have never been before might be the secret to taking on established competitors in the financial technology sector.
As a result, Grok is also under EU DSA surveillance. The EU is in “very close contact with X on the launch of Grok,” according to a senior Commission source who said this today.
Without revealing precisely which aspects have been blocked, the official said X has postponed the introduction of several Grok features in the area until after the impending European Parliament election.
“X has postponed the release of a portion of the Grok feature until after the election,” the spokesperson informed reporters. The spokesperson stated, “I believe this is an acknowledgement on their part that certain characteristics could be risky in the setting of civic discourse and elections, given the continuous investment.”
We inquired of the Commission as to which Grok features it thought X had frozen in the EU. “We observed the phased deployment of Grok in the EU, with a first version only visible and accessible to X Premium subscribers and not the wider public,” an official stated.
“We reserve our position on Grok’s suitability for the DSA requirements,” they said.
We also contacted X about Grok’s EU debut, but it had not replied by press time.
Grok is an X premium feature by default, so it’s unclear what feature compromises were made for the AI chatbot’s EU debut.
The Commission spokesman provided more comprehensive remarks on generative AI chatbots integrated into a designated VLOP, or very large online search engine (VLOSE). He said that platforms are legally required to carefully consider risks and put in place workable solutions to them, such as protections against so-called “hallucinations,” which happen when these GenAI tools make up information and present it as fact.
The spokesman emphasised that the DSA election rules have recently reaffirmed the requirement for ex ante risk assessment for “important features” like GenAI assistants, an area where the Commission launched a recent inquiry ahead of upcoming European elections.
In March, 8 VLOP/SEs, including X, received ad hoc information requests about risk assessment and mitigation strategies for risks arising from generative AI-produced content.
