South Wales Police has reopened its investigation into sexualised deepfakes of Jess Davies made using Grok, according to an October 22, 2026 report by The Bureau of Investigative Journalism. The newsroom says it supplied officers with information about the account involved, including a cryptocurrency wallet and dozens of deleted posts.
Police are now investigating under sexual-offence legislation designed to address deepfakes and following up enquiries with X, TBIJ reports. The earlier investigation had closed in May because officers could not identify the user.
The reopening concerns one named woman’s complaint. It is an ongoing investigation, with no criminal finding made. Two documented changes stand out: investigators have additional account information to examine, and police are approaching the alleged conduct under a different legal framework.
How Davies’ Complaint Reached a Second Investigation
Davies, a presenter and campaigner against deepfake abuse, reported the image to South Wales Police on January 2, 2026. According to TBIJ’s September 29 investigation, an anonymous account had asked Grok to turn her profile photograph into a sexualised image, which the chatbot then posted publicly.
The September report said Davies was not invited to give a statement until March. Police subsequently closed the investigation without identifying the person behind the account, even though the account remained active.
The published reporting gives this sequence:
| Date | Documented development |
|---|---|
| January 2, 2026 | Davies reports the post to South Wales Police. |
| March 2026 | Davies is invited to give a statement. |
| May 2026 | The investigation closes after officers cannot identify the user. |
| September 2026 | Davies files a formal complaint; TBIJ publishes its investigation on September 29. |
| October 22, 2026 | TBIJ reports that the case has reopened following its reporting and provision of account information. |
The January and March dates come from TBIJ’s September investigation; the closure month, formal complaint and reopening are documented in its October update.
Davies welcomed the decision, telling TBIJ: “I am pleased that the decision has been made to reopen my case”.
She also expressed frustration at having to push for a review. Reopening gives police another opportunity to investigate, but it is not itself a resolution of her complaint.
What the Wallet and Deleted Posts Could Add
TBIJ says the information it gave South Wales Police included a cryptocurrency wallet and dozens of deleted posts associated with the user who targeted Davies. Those details are the newsroom’s account of its own work; they have not been independently verified for this article.
The potential value is account attribution: connecting an online account to the person operating it. TBIJ’s earlier reporting identified that as a central obstacle in Davies’ case. The account was visible, but officers had not identified its operator.
A cryptocurrency wallet can provide another identifier for investigators to examine. It does not automatically provide a person’s name. Deleted posts can potentially supply context or links between online activity, but the update does not describe their contents, how TBIJ recovered them, or what connections police have established.
The published evidence therefore supports a narrow conclusion: police received additional leads before reopening the case. It does not show that the wallet has been traced to a named suspect or that the deleted posts prove who made the request.
My view is that this is the most consequential detail in the update. A complaint previously closed over an identification problem is being reconsidered after a third party supplied account information. That makes the usefulness of those leads a more concrete issue than general claims about whether an AI platform has improved its safeguards.
Why the Legal Reframing Matters
South Wales Police now says it is investigating under sexual-offence legislation designed to tackle deepfakes, according to TBIJ’s October report. That update describes the earlier handling as having been under malicious communications law.
The September account was less definite about the original legal basis. Davies said officers had discussed malicious communications legislation, but TBIJ reported that the force had not clarified which law it had used. The October update provides the clearer account of the change.
The timing of the alleged conduct is important. TBIJ reported that a new law making it illegal to create, or request the creation of, an adult’s intimate image without consent took effect on February 6, 2026. Davies had reported the post more than a month earlier.
However, TBIJ also reported that distributing a non-consensual intimate deepfake was already a crime when she complained. That separates two questions:
- Creation or requesting creation: which rules applied to producing the image at the relevant time.
- Distribution: which rules applied to making the resulting image available to others.
Grok’s public posting mechanism complicates the second question. According to Davies’ account in the September investigation, police had raised a possible issue over whether Grok, rather than the individual requester, could be treated as responsible for sharing the image.
That was a concern described in the reporting, not a court ruling. The October update does not identify a particular charge or explain how investigators now intend to address responsibility for the automated post.
The practical implication is that the chatbot’s role cannot be understood solely as an image-generation feature. In this complaint, the request also led to a public post. Selecting the applicable offence requires attention to that sequence and to the law in force at the time. Pointing only to the later creation offence would leave out the earlier distribution question.
X’s UK Restrictions Address a Different Problem
TBIJ’s October report says that, according to Ofcom, UK users have been blocked from viewing, generating or sharing pornographic content through Grok’s standalone image-generation tool.
The scope is specific: UK users, pornographic content and the standalone image tool. It should not be read as a worldwide restriction or a guarantee that every form of non-consensual sexualised editing has been prevented.
The same report recounts TBIJ’s finding that, despite restrictions introduced by X, users had still been able to request bikini images and other sexualised deepfakes of women without their consent. It also describes edits to images of Davies through the standalone Grok app. TBIJ says one image was subsequently removed from X after the newsroom flagged it to her.
These details concern different kinds of content and different platform actions. A restriction on pornographic generation, removal of a particular post and an investigation into an earlier user’s conduct each address a separate part of the problem.
For Davies’ reopened case, the distinction is practical. Blocking future access to certain content does not identify the person behind an earlier account. Removing an image also leaves the question of criminal responsibility unresolved. Police enquiries with X may be relevant to that identification work, but the update does not say what information the platform has provided.
Frequently Asked Questions
3 questions
1Why Does February 6 Matter?
TBIJ reports that the offence covering non-consensual creation or requests for adult intimate images took effect on February 6, 2026. Davies’ January complaint predates it, although distributing non-consensual intimate deepfakes was already criminalised.
