The former head of trust and safety at Twitter has warned the platform now known as X is charting a collision course with the European Union’s rebooted digital rulebook, the Digital Services Act (DSA) — which carries penalties of up to 6% of global annual turnover for confirmed breaches of the online governance regime.
Speaking during an on-stage interview at the Code Conference this week, Yoel Roth pointed to X’s decision back in May to withdraw from the EU’s Code of Practice on Disinformation, as well as citing recent remarks by commissioner Vera Jourova which singled out the platform as the worst for the spread of disinformation — predicting a clash with the bloc’s regulators is now “inevitable”.
“Regulatory time moves a lot slower than internet time. And so I think we’re going to see lagging effects here. But it is inevitable,” he suggested. “If the European Union has proven anything, it’s that they are willing and able to regulate large companies and push them to abide by the laws of the European Union. And so if I had to make a prediction, it would be that it won’t be right now — it might not even be a year from now — but there will be consequences.
“The question is how much damage happens between now and then to individual people who work at companies — like me — to the quality of the conversation on Twitter, to the platform itself? I think there’s a lot that can be done before regulation catches up with reality and that’s what really worries me.”
X marks the odd one out
Remembering what he was thinking when he left Twitter last year, a short time after Elon Musk’s acquisition ushered in a new era of platform drama, Roth said he had believed that commercial and regulatory factors would act as a constraint on what the new owner might do as regards damage to trust and safety.
But he said his assumptions have turned out to be wrong — citing the exodus of advertisers Musk has presided over and the decision to pull the platform out of the EU’s disinformation Code. “They’re the only large platform to do so,” he noted. “And just today, Commissioner Jourova has said that Twitter are tempting fate — that they are an easy target for enforcement. And so alright, I was I was wrong on that point as well.”
Back in April, just prior to Musk pulling X out of the Code, the EU designated the platform a so-called VLOP (very large online platform) under the DSA — which means it has a legal requirement to tackle systemic threats such as disinformation.
The bloc has also made it clear that its regulators will treat adherence with the (non-legally binding) Disinformation Code as a signal it factors in when assessing whether larger platforms are in compliance with the (legally binding) DSA.
Neglecting the societal threats posed by disinformation thus risks major sanction in the EU. And not just financial; as well as big fines, the DSA empowers the Commission to block services that repeatedly fail to comply with the rulebook — so there is the additional possibility, if X entrenches itself on a reckless trajectory and is unable to demonstrate it’s tackling safety issues, that it could ultimately lose access to the EU market.
In further worrying signals for election security at Twitter, The Information reported earlier this week that Musk had cut half the remaining members of the election integrity team — flying in the face of claims by the company, and recently repeated by its CEO Linda Yaccarino, that it’s expanding efforts to tackle threats to elections.
Also this week it emerged the platform has quietly removed a legacy option to report misleading information about politics — although, as we noted in our coverage, X users in the EU can still find an option to report “negative effects on civic discourse or elections” (under a region-specific option to “report EU illegal content”). So it has not blocked EU users’ ability to report election integrity concerns entirely.
However, without enough staff internally to handle reports it’s clear X isn’t going to be able to effectively tackle political disinformation or get visibility into developing threats to elections — including in the EU.
Under Musk, the platform’s own “Civic Integrity” policy does technically prohibit “manipulating or interfering in elections or other civic processes” but as we know from how things went down at legacy Twitter for years — before the former leadership’s extremely gradual conversion to “conversational health” — having a policy and enforcing a policy are two very different things.
Election integrity threat
Roth raised the fate of the election integrity team during the Code interview. “The last remaining staffers at Twitter who have expertise in election security, were all summarily fired,” he said. “When I went into this year, and people started asking me what do you predict about 2024, I was like, look, there’s one person still left at Twitter in Dublin, who has single handedly kept global elections from going off the rails. I hired him myself. He’s brilliant. For as long as he can stick it out there’s a hope. [But he’s been] summarily fired so I’m not super optimistic about that.”
The employee he mentioned is Aaron Rodericks, X’s Dublin-based co-lead on trust & safety and threat disruption.
Per a report in Rolling Stone, Rodericks was targeted for harassment by right-wing influencers following a jobs post he made on LinkedIn advertising the fact he was looking to recruit eight more staffers to the election integrity team at X.
A series of online attacks, by accounts including Mike Benz‘s, accused the X staffer of being a “censorship shill”; suggested he was trying to recruit CIA operatives to work at X; and claimed he’d liked tweets that were critical of Musk — soon after which the company instigated disciplinary proceedings against him. Then, last week, Irish press reported Rodericks had secured a temporary High Court injunction. His account of X’s actions as a “complete sham” persuaded the court to grant an injunction temporarily suspending the disciplinary proceeding.
We’ve confirmed Rodericks remains technically an X employee but the legal dispute is ongoing. And — clearly — he is not in a position to carry out the work he was doing internally on election integrity before a handful of right-wing trolls began a campaign of targeted harassment in a bid to push Musk’s buttons and get him fired.
A realization that Musk was taking arbitrary decisions rather than following “the rule of law” was the final straw for Roth last year, when he took the decision to leave. “All of this reads to me like a company that has abandoned the rule of law,” he said during the interview. “Not just the laws of the land, like the Digital Services Act, but also the laws that it imposed on itself — the operating principles that guided the company.”
No trust
Roth also revealed that death threats he received after leaving Twitter last year have not been removed from the platform — undermining counter claims being made by CEO Linda Yaccarino that X is making strides to improve safety.
“I would encourage Twitter to take a look at the death threats targeting me,” said Roth. “The death threats that were inspired by the company’s leader. They’re all still there. Twitter didn’t take them down. Thousands of them. They’re still on the platform today.”
During the interview Roth also made short shrift of a study X put out back in May, in partnership with a third party company called Sprinklr, which had claimed there is a very low prevalence of English language hate speech on the platform — saying the claims are “completely non auditable” so simply cannot be trusted.
Notably under Musk’s leadership X has also made it far harder and more expensive for researchers to access data to conduct independent studies — despite the DSA placing requirements on larger platforms to support public interest research into algorithmic effects. So, again, he’s moving counter to the direction of travel EU regulators are demanding with their digital rulebook.
“By any measure [safety on Twitter] is worse, except by Twitter’s measure,” argued Roth. “We have seen, just this week, a study out from researchers in Europe talking about the prevalence and spread of disinformation across all of the major platforms. I will give you one guess which platform has the highest degree of spread: It’s Twitter. We have also seen research that suggests that the prevalence of hate speech and abuse on the platform is higher. We’ve seen independent research that suggests that ISIS has staged a 70% return on Twitter. This isn’t like free speech. This is ISIS, right? Like we’re not talking about the grey areas of content moderation.”
“Peer review is a pain in the ass. Every academic will tell you that. But the reason that it exists is so you can answer these questions in a satisfying empirical way. You can say if we’re talking about hate speech, it is defined in a rigorous way. We don’t know that about Twitter’s data,” he added. “There’s simply no way to know. You don’t.”
Liquidating the company’s comms team was another early Musk Twitter ‘reform’. Since then, the platform has typically ignored press requests for comment and/or sent a meaningless auto-reply. (Its latest auto-response reads: “Busy now, please check back later.”) But CEO Yaccarino — who was interviewed at the Code conference shortly after Roth — rejected Roth’s assessment, repeating the claim that safety on X has improved since Musk took over.
Also, in a slight departure from the usual X stone-walling of press enquiries, when we emailed asking for a response to Roth’s concerns, in addition to aforementioned auto-response, we got a (brief) reply from Joe Benarroch, a former NBCUniversal exec (and former colleague of Yaccarino there) who she convinced to join her at her new employer this summer. “Linda addressed this on stage at Code,” he suggested, adding: “If memory serves me correctly, a lot of what Yoel discussed was similar to the points he made at Code a year ago.”