Following months of pressure that intensified during the fall and winter, Britain’s communications regulator said on Friday that X had agreed to a set of obligations on illegal hate speech and terrorist content with Ofcom.
According to the agreement, Elon Musk’s platform will evaluate at least 85% of suspected illegal hate and terrorism posts within 48 hours, analyze them within 24 hours on average, and provide the regulator with quarterly performance data over the course of the next year.
Additionally, the platform has pledged to limit UK access to accounts run by or on behalf of organizations prohibited by British terrorism law and to hire outside experts to improve a reporting system that civil society organizations have frequently called opaque.
The language is important because the majority of complaints against X that have been lodged with Ofcom over the past year have focused on flagged content that has not been properly received or addressed.
Oliver Griffiths, director of Ofcom’s online safety group, said:
“We have evidence that terrorist content and illegal hate speech is persisting on some of the largest social media sites,” “This is of particular importance in the UK following a number of recent hate-motivated crimes suffered by the country’s Jewish community.”
In a statement, Suzanne Cater, the director of online safety enforcement at Ofcom, stated that “terrorist content and illegal hate speech is persisting on some of the largest social media sites” and that the disparity has become “of particular importance in the UK following a number of recent hate-motivated crimes suffered by the country’s Jewish community.”
The pledges came after “sustained campaigning” following the attack on the Heaton Park Synagogue outside Manchester last year, according to Imran Ahmed of the Center for Countering Digital Hate.
Britain has had a challenging period of events to deal with. A deadly incident in north London last month that police are examining as terrorism followed the Heaton Park attack, and CCDH’s own monitoring following the Golders Green attack revealed what it called a deluge of antisemitic posts on X (the underlying CCDH dataset is here).
Those incidents are not specifically addressed by the new promises. The procedural floor is positioned beneath them. There was a mixed response. The package was “a good start,” according to Antisemitism Policy Trust CEO Danny Stone, but X was still “failing in so many regards” to address racism.
Ofcom was cautious to point out that its official investigation into X, which includes the company’s procedures for dealing with illicit content and queries posed by its Grok AI assistant, is still ongoing. The agreement reached Friday is not a settlement, but a negotiated commitment.
Another track by Grok is playing at the same time. X this month limited Grok’s image editing to paid users after a deepfake scandal and a threat of a ban in the UK, where the regulator Ofcom is investigating X’s treatment of AI-generated sexualized images created using the chatbot. The Friday obligations don’t address that thread. They sit on the edge of it.
The broader context is familiar to anyone who follows the platform’s regulatory pipeline. The European Commission’s own monitoring shows that X is the biggest source of misinformation, and the Commission is now investigating whether the firm is doing enough to curb hate speech. Regulators in Singapore and Australia have dealt with related issues. The UK agreement ends up in the middle, rather than at the back of a queue.
The new commitments are essentially an expression of the Online Safety Act framework, which came into effect in 2023, and requires the biggest platforms to remove any illegal content quickly or face fines of up to 10% of their global turnover.
On paper, the 24-hour review vow is the kind of quantifiable metric that the regulator has desired. The 85%-within-48-hour backstop appears to be calculated so that Ofcom can conduct an audit.
Delivered over the course of the next year, the quarterly statistics will be the regulator’s first detailed dataset on whether platform-side promises genuinely shift the removal of illegal content in the direction that the law intended.
