Liz Kendall’s response to X ‘nudification’ is nice – however not sufficient to resolve the issue

[ad_1] O n X, a lady posts a photograph in a sari, and inside minutes, varied customers are beneath the submit tagging Grok to strip her right down to a bikini. It's a stunning violation of privateness, however now a well-recognized and commonplace apply. Between June 2025 and January 2026, I documented 565 situations of customers requesting Grok to create nonconsensual intimate imagery. Of those, 389 had been requested in simply sooner or later.

Final Friday, after a backlash towards the platform’s capability to create such nonconsensual sexual photographs, X introduced that Grok’s AI picture era characteristic would solely be accessible to subscribers. Experiences counsel that the bot now not responds to prompts to generate photographs of ladies in bikinis (though apparently will nonetheless achieve this for requests about males).

However because the know-how secretary, Liz Kendall, rightly states, this motion “doesn't go wherever close to far sufficient”. Kendall has introduced that creating nonconsensual intimate photographs will change into a felony offence this week, and that she is going to criminalise the availability of nudification apps. That is acceptable, given X’s weak response. Putting the characteristic behind a paywall implies that the platform can extra immediately revenue off the web dehumanisation and sexual harassment of ladies and minors. And stopping the “bikini” responses after public censure and the specter of laws is the least X can do – the larger query is why it was even doable within the first place.

These measures are a step ahead. The shadow know-how secretary, Julia Lopez, recommended in her response that the federal government was overreacting, that this was simply “a modern-day iteration of an outdated drawback”, no completely different from crude drawings or Photoshop. She’s fallacious. The dimensions is completely different. The accessibility is completely different. The pace is completely different. With Photoshop, there's a technical ability required, in addition to direct publication by a person, which locations all actions besides platform provision squarely on them. On this case, although, the person makes a daily textual content reply with a request, and Grok generates and publishes felony abuse to an enormous viewers.

Kendall’s strategy criminalises customers who create or alter these photographs, and the businesses that offer devoted nudification instruments. That’s the place it misses the purpose. Grok and most outstanding image-generation instruments will not be devoted nudification instruments. They're general-purpose AI with weak safeguards. Kendall will not be asking platforms to implement proactive detection. The regulation waits for hurt to occur, then punishes.

The drawbacks of this strategy are apparent. I noticed this materials being generated for months earlier than the mainstream backlash started. These are dangerous photographs that had been generated that also exist, and maybe had been saved and shared throughout different platforms. For the victims of this AI sexual abuse materials, regulation after the actual fact received’t assist. For hurt that's structurally amplified on this method, the strategy have to be preventive, not reactionary.

One other extra basic drawback is that whereas the UK pushes AI security regulation, the US is transferring in the wrong way. The Trump administration desires to “improve america’ world AI dominance by means of a minimally burdensome nationwide coverage framework for AI”. Beneath this framework, there may be little incentive for American AI firms to control misuse of their merchandise. This issues as a result of AI regulation is incomplete with out cross-border collaboration. Kendall can criminalise customers within the UK, she will threaten to ban X totally. However she can't cease Grok from being programmed in San Francisco. She can't pressure OpenAI or Anthropic or another US firm to prioritise security over pace. With out US cooperation, we try to control a transnational know-how with nationwide legal guidelines.

Whereas this wrangling over regulation and updating coverage performs out, many victims, and different girls on-line, will likely be questioning what this new period of AI-enabled on-line sexual harassment means for them, and questioning their participation on world social media platforms. If my picture has been digitally altered, how do I get justice if the perpetrator is midway internationally? Transparency within the practices of AI firms is in decline – so how may the identical firms be trusted to be accountable and audit techniques that reproduce hurt?

The reality is that these firms can't be trusted. That is why globally, regulation must shift from “take away hurt if you discover it” to “show that your system prevents hurt”. We should code energy into the method by requiring obligatory enter filtering, impartial audits and licensing circumstances that make prevention a authorized technical requirement. This implies it might catch the hurt earlier than it materialises, enabling regulation to minimise dangerous behaviour by these AI firms earlier than their merchandise are deployed. That is the kind of work that we on the AI Accountability Lab within the Adapt Centre at Trinity Faculty Dublin are pushing ahead by means of our analysis.

Regulation after the actual fact is best than nothing. Nevertheless, it gives little to the victims who've already been harmed, and sidesteps the conspicuous absence of regulation enforcement in addressing these platform harms.