Keeping up with AI and social media: Assessing Northern Ireland’s Move to Criminalise Sexually Explicit Adult “Deepfakes”

Graeme Littlemore

“Our collective failure to act with any urgency over the easy creation and distribution of sexual digital forgeries has created a culture in which women are now living with the ever-present ‘invisible’ threat of deepfake sexual abuse”, McGlynn and Toparlak state in their critical article on the “new voyeurism”.

As it stands, there is no legislation in Northern Ireland which criminalises this behaviour where images relate to adults (see paragraph 1.8 of the public consultation on this proposed legislation). Northern Ireland’s Minister of Justice has promised to introduce such amendments to the Justice Bill at consideration stage of the Justice Committee in the Northern Ireland Assembly. A further amendment has been announced by an Alliance Party MLA to ban “having and creating the technology", seeking to target the technology underpinning this behaviour. 

At the time of writing, X is under international scrutiny for its AI bot “Grok”, which has been used to create non-consensual sexualized images. Women and girls generally make up 99% of those targeted with the vast majority of those creating these images being men. X announced changes to its settings so that “image generation and editing are currently limited to paying subscribers” and issued warnings to users. Ofcom, the UK’s regulator for communications industries, has promised to investigate. The UK Government is having conversations on its continued use and Elon Musk has repliedthat “They just want to suppress free speech”. Indonesia recently became the first country to temporarily block this tool.

This article focuses on assessing and situating Northern Ireland’s proposed legislation within the corpus of other related UK legislation. Firstly, the proposed amendments will be outlined. Secondly, definitional issues will be discussed. Thirdly, legislative overcrowding will be discussed. Lastly, jurisdictional protection gaps will be outlined.

The Proposed Legislative Amendments

Northern Ireland’s Department of Justice is proposing to criminalise four different offending behaviours that would be dealt with as hybrid offences in line with current Northern Ireland voyeurism offences related to images of adults (see paragraphs 4.1, 4.2 and 4.4 of the consultation).

This includes:

1.     Intentionally creating a sexually explicit deepfake image of an adult, without consent, or a reasonable belief in consent, with the intention of causing humiliation, alarm or distress to the person depicted in the image.

2.     Intentionally creating a sexually explicit deepfake image of an adult, without consent, or a reasonable belief in consent, for the purposes of sexual gratification.

3.     Intentionally requesting the creation of a sexually explicit deepfake image, without consent, or a reasonable belief in consent, with the intention of causing humiliation, alarm or distress to the person depicted in the image.

4.     Intentionally requesting the creation of a sexually explicit deepfake image, without consent, or a reasonable belief in consent, for the purposes of sexual gratification.

The Department stipulated that “the motivation of causing humiliation, alarm or distress and the sexual gratification motivation are proposed for the offences” (see paragraph 4.5 of the consultation). To guard against over-criminalisation, a recklessness element is proposed (see paragraphs 3.11 to 3.15 of the consultation). Perpetrators convicted of any of the offences (only where sexual gratification is proven) will be subject to appropriate sex offender notification requirements (see paragraph 3.19 of the consultation).

Three additional offending behaviours are proposed related to sharing or threatening to share a sexually explicit deepfake image but are not dealt with directly in this article (see paragraph 5.1 of the consultation).

However, constructing a modern definition of deepfakes that captures the gendered, victim-centered, and quickly evolving landscape of AI remains a continued difficulty in this area of law.

Definitional Issues and Futureproofing Legislation

The Department did not suggest a definition of sexually explicit deepfake images at this stage but noted it must keep pace with technological advancements (see paragraph 7.9 of the consultation). Ofcom defines deepfakes as “audio-visual content that have been generated or manipulated using AI, which misrepresent someone or something” with intent to harm and elements of deception. Women’s Policy Group NI in response to these proposals did not prefer ‘deepfake’ and instead suggested “intimate digital forgeries”. This corresponds with McGlynn and Toparlak’s article which states that “deepfake” is rooted in an online community that develops non-consensual pornography and holds connotations with consensual adult material. The Northern Ireland Human Rights Commission (NIHRC) suggested that a definition will need to be clear enough, victim-centered, and able to rapidly respond to technological advancements.

The current controversy over X is a test of this proposed law’s flexibility to rapidly respond to new ways of offending and raises concerns as to whether a future definition can keep up with technology. It was acknowledged that further planned amendments to criminalise deepfakes in Northern Ireland would not be applicable to X as its AI software is built into the social media site itself. The effectiveness of addressing technological root causes in Northern Ireland remains to be seen, particularly as internet services are a reserved matter under Schedule 3 of the Northern Ireland Act 1998.

Overcrowding the Legislative Field

It is well observed that, in the absence of standalone legislation, those impacted will continue to face a complex patchwork of existing laws with different regulatory focuses. A 2022 briefing paper noted that Northern Ireland previously criminalised revenge pornography under section 51 of the Justice Act (Northern Ireland) 2016 relating to disclosing private sexual photographs and films with intent to cause distress with similar reasonable belief provisions. Importantly, it was acknowledged that it was unclear whether “deepfake revenge pornography” would fall within the umbrella of this offence. Overlaps with existing offences related to “cyberflashing and deepfake pornography” were notedin Northern Ireland with relation to, but not limited to, the The Protection from Harassment (Northern Ireland) Order 1997, where section 6 provides for conduct that puts others in fear of violence. Moreover, section 2 of the Justice (Sexual Offences and Trafficking Victims) Act (NI) 2022 deals with the intentional sending of an unwanted image without reasonable belief of consent. Further, the Malicious Communications (NI) Order 1988 makes an offence the sending of letters and articles with intent to cause distress or anxiety.

While proposed amendments in Northern Ireland are certainly welcomed, they risk becoming situated amongst a complex, overlapping and crowded legislative and regulatory landscape which may contain unintended legal protection gaps rendering them less effective (see paragraph 3.37 of the NIHRC response).

Confronting Jurisdictional Protection Gaps in Legislation

Legal protection gaps in Northern Ireland’s proposed amendments are contrasted with legislation in England and Wales under the Data (Use and Access) Act 2025. Section 66F(6)(c) relates to requesting the creation of purported intimate image of adults regardless of where in the world the perpetrator is located. In their written evidence to a Westminster Parliamentary committee on the UK Government’s previous plan to criminalise the creation of sexually explicit deepfakes, McGlynn and Davies noted significant gaps in the law related to solicitation cases. They noted that most cases of solicitation will not be covered by existing law (as it related to England and Wales at the time) as the creator is anonymous, and even if their location is known, they likely reside in a jurisdiction that does not criminalise this behaviour. They further stipulated that these solicitation offences must include specific provisions on the extra-territorial effect of the offence.

Notably, current proposals in Northern Ireland are silent on this extra-territoriality element, reflecting UK-wide imbalances to meet the wide-reaching nature of AI and social media. Additionally, Westminster's Science, Innovation and Technology Committee wrote to the Government expressing concern over this current X controversy, stipulating that the UK’s Online Safety Act 2023 is “riddled with gaps – including its failure to explicitly regulate generative AI”. This reinforces arguments by Women’s Policy Group NI to implement harmonized legislation to achieve greater clarity, effectiveness, and fill legal gaps that the NIHRC cautioned against. 

Concluding Remarks

The latest controversy over X represents another rapidly evolving way that unscrupulous actors will find new ways of offending. Newly introduced legislation must be gendered, victim-centered, and sufficiently clear, and must keep pace with these technological advancements. Cross-jurisdictional legislative gaps must be addressed, and consideration should be given to standalone legislation in order to harmonize an increasingly overlapping and crowded legal field and protect those most impacted.