Changes to Elon Musk’s AI Grok ‘insulting’ to victims, says No 10

The UK authorities says Elon Musk’s platform X limiting Grok AI picture edits to paid customers is “insulting” to victims of misogyny and sexual violence.

Talking on Friday, Downing Road mentioned the transfer “merely turns an AI function that permits the creation of illegal photos right into a premium service”.

It follows vital backlash after Grok digitally altered photos of others by undressing them – one thing it says it now can solely do for many who pay a month-to-month payment.

Nevertheless, it has no such restrictions when accessed by its “edit picture” performance, separate app or web site.

The BBC has approached X for remark.

The prime minister’s official spokesperson advised reporters on Friday it confirmed X “can transfer swiftly when it desires to take action”.

They mentioned it was “abundantly clear that X must act and must act now”.

“It’s time for X to grip this situation, if one other media firm had billboards on the town centres displaying illegal photos, it could act instantly to take them down or face public backlash,” they added.

The Liberal Democrats have known as for entry to X to be quickly restricted within the UK whereas the social media website is investigated.

“Musk has thrown his toys out of the pram in protest at being held to account for the tsunami of abuse,” mentioned Professor Clare McGlynn, an skilled within the authorized regulation of pornography, sexual violence and on-line abuse.

“As an alternative of taking the accountable steps to make sure Grok couldn’t be used for abusive functions, it has withdrawn entry for the overwhelming majority of customers.”

And Hannah Swirsky, head of coverage on the Web Watch Basis, mentioned it “doesn’t undo the hurt which has been completed”.

“We don’t consider it’s ok to easily restrict entry to a device which ought to by no means have had the capability to create the type of imagery now we have seen in current days,” she mentioned.

The charity previously said its analysts had found “prison imagery” of ladies aged between 11 and 13 which “appeared to have been created” utilizing Grok.

“Sitting and ready for unsafe merchandise to be abused earlier than taking motion is unacceptable,” Swirsky mentioned.

Grok is a free device which customers can tag instantly in posts or replies underneath different customers’ posts to ask it for a specific response.

However the function has additionally allowed individuals to request it to edit photos – and ask it to digitally strip individuals of most of their clothes.

Grok has fulfilled many person requests asking it to edit photos of girls to point out them in bikinis or little clothes – one thing these topic to such requests have advised the BBC left them feeling “humiliated” and “dehumanised“.

Nevertheless as of Friday morning, Grok has advised customers asking it to change photos uploaded to X that “picture era and modifying are at the moment restricted to paying subscribers”.

It provides customers “can subscribe to unlock these options”.

Some posts on the platform seen by BBC Information counsel solely these with a blue tick “verified” mark – unique to X’s paid subscriber tier – have been in a position to efficiently request picture edits to Grok.

Dr Daisy Dixon, a lecturer in philosophy at Cardiff College and feminine X person who mentioned she had seen a rise in individuals utilizing Grok to undress her, welcomed the change however mentioned it felt “like a sticking plaster”.

“Grok must be completely redesigned and have built-in moral guardrails to forestall this from ever occurring once more,” she advised the BBC.

“Elon Musk additionally must acknowledge this for what it’s – one more occasion of gender-based violation.”

Charities campaigning to finish violence in direction of ladies additionally criticised X over the transfer.

“To limit entry to paying subscribers just isn’t solely insufficient – it represents the monetisation of abuse,” mentioned Emma Pickering of Refuge.

She mentioned whereas limiting the function to paid X customers could scale back quantity, abuse had “merely been positioned behind a paywall, permitting X to revenue from hurt”.

In the meantime Andrea Simon, director of the Finish Violence Towards Ladies coalition, mentioned “we’re not satisfied that X will take motion to establish and forestall this, given it has dragged its heels with tackling this abuse”.

Prof McGlynn mentioned X’s response echoed its strategy to pornographic Taylor Swift deepfakes on the platform last year – the place it blocked searches for sexualised materials generated of the popstar utilizing a Grok AI video function.

“He’s doing this to stoke free speech arguments,” she added.

“He’ll declare regulation is stifling individuals’s use of this expertise. However, all of the regulation requires is that he takes crucial precautions to scale back hurt.”

Ofcom’s powers underneath the Online Safety Act embrace having the ability to search a court docket order to forestall third events from serving to the Elon Musk-owned platform elevate cash or be accessed within the UK.

The BBC has approached the regulator for remark.

Lib Dem frontbenchers Victoria Collins, Marie Goldman and Max Wilkinson have written to Ofcom boss Dame Melanie Dawes, asking her to take motion.

“If some other platform have been discovered to be facilitating the mass era of sexual abuse imagery at this scale, it’s tough to consider it could be permitted to proceed working uninterrupted,” they wrote.

“Ofcom should prohibit entry to X now whereas an investigation takes place.”

Downing Road reiterated the federal government’s “full help” for regulator Ofcom to use all its powers – as much as and together with a ban.

Prime Minister Sir Keir Starmer beforehand known as photos generated by Grok “disgraceful” and “disgusting”.

In the meantime the Conservatives reiterated feedback made following PMQs on Wednesday by a spokesperson for Kemi Badenoch, who known as deepfake photos “completely abhorrent”.