Meta 'too slow' during UK riots to deal with violent posts online, says independent review (2025)

Meta was too slow to act during the UK's riots last summer, according to the US tech giant's independent review body.

After three young girls were murdered in Southport at a dance class on 29 July 2024, misinformation and disinformation spread on social media falsely suggesting the perpetrator was a Muslim and an asylum seeker, said the new report.

People on online platforms encouraged violence and rioting across the country and the disorder lasted for nine days between 30 July and 7 August.

Police arrested more than 1,280 people for their involvement in the unrest, according to the National Police Chiefs' Council.

Now, Meta's oversight board says it had "strong concerns" about the company's ability to moderate hateful and violent imagery in the report.

It said the social media giant, which owns Facebook, Instagram and WhatsApp, was "too slow" to implement an emergency crisis protocol to help with moderation, and raised concerns about Meta chief executive Mark Zuckerberg's "hastily announced" policy changes that saw the company move away from using third-party fact-checkers in January.

"We don't know enough," Paolo Carozza, co-chair of the oversight board, told Sky News.

"Meta needs to be very careful and deliberate about studying and verifying the effectiveness of a [community notes system]."

The oversight board reviewed three posts shared during the UK riots, all of which were left up by Meta's automatic moderation systems despite being flagged by users.

The first called for mosques to be smashed and buildings where "migrants", "terrorists" and "scum" live to be set on fire, which the oversight board said was a serious violation of Meta's policies.

"There is no way to interpret this post as a casual or non-serious statement," said the report.

The second post showed an AI-generated image of a giant man in a Union Jack t-shirt chasing smaller Muslim men, with text over the image giving details of a protest to attend during the riots. It was accompanied by the hashtag "EnoughisEnough".

"[The post] is a clear call for people to gather and carry out acts of discriminatory violence," concluded the report. It added Meta's conclusion that the post contained no target or a threat "strains credibility".

Read more from Sky News:
Country star seeks pardon from criminal past
What are the chances of a British pope?

The third post showed an AI-generated image of four Muslim men running through Westminster after a crying blond-haired toddler. In the image, one of the men waves a knife and a plane flies towards Big Ben.

Meta didn't remove this image as it said it referred to the specific Muslim man who was incorrectly accused of the children's murders in Stockport. Referring to the incorrectly identified man was allowed under Meta's policies.

The oversight board said it "does not depict the Southport stabbing in any form" and ruled it should have been removed.

"In response to these events last summer, we immediately set up a dedicated task force that worked in real time to identify and remove thousands of pieces of content that broke our rules - including threats of violence and links to external sites being used to coordinate rioting," a Meta spokesperson told Sky News.

They said Meta "will act to comply with the board's decision".

Read more from Sky News:
Blood test that can detect 12 common cancers to be trialled on NHS
MMA fighter jailed for 15 years

The UK riot posts were part of a number of posts examined by the oversight board to assess Meta's content moderation.

In contrast to the three Southport-related posts, which it recommended be removed, the board recommended Meta keep online two posts that may have been "highly offensive" to the trans community.

Get Sky News on WhatsApp

Follow our channel and never miss an update

Tap here to follow

"Protection of freedom of expression doesn't protect an immediate incitement to violence, but it does protect controversial and even offensive speech," said Mr Carozza.

"A lot of what needs to be done, both by the oversight board and Meta, is weigh where we are most willing to allow the errors to fall. We want to err on the side of more inclusivity of speech."