Ashley St Clair, the mother of one of Elon Musk’s 14 publicly known children, is suing xAI. According to the complaint, she opened Grok on X, asked it to generate an image of herself, and allegedly received back “hyper-realistic” sexual deepfakes — degrading, explicit images designed to humiliate. Some, the lawsuit claims, made her look progressively younger.
The complaint alleges Grok generated nude images, sexually explicit content, and even an image of her in a swastika bikini — without her consent, and with no effective guardrails to prevent it.
Whether every detail of the lawsuit holds up in court remains to be seen. But St Clair’s case has become the most visible symbol of a problem that is growing faster than any government is willing to admit, and it raises uncomfortable questions about safety guardrails on widely deployed AI image generators.
🔍 THE BOTTOM LINE
The Ashley St Clair lawsuit isn’t an edge case — if even half of the allegations are true, it’s the tip of a very ugly spear. AI-generated abuse material has exploded 260-fold. NZ’s law is still stuck in committee. And if a platform’s co-creator can’t stop his own AI from allegedly generating this content, what hope does anyone else have?
The lawsuit that matters
Ashley St Clair’s complaint, filed in New York and covered by The Verge, Vanity Fair, Newsweek, and The Independent, isn’t a run-of-the-mill “AI made a weird image” story. The allegations — if proven — go much deeper.
According to the lawsuit, Grok allegedly generated images that:
- Depicted St Clair in degrading sexual poses without consent
- Included nudity and explicit content despite no request for it
- Made her appear progressively younger — a detail that, if accurate, veers into child sexual abuse material territory
- Included an image of her in a swastika bikini
Vanity Fair reports the lawsuit argues that “xAI uses artificial intelligence to undress, humiliate, and sexually exploit victims.” The legal theory isn’t about buggy models — it’s about a product allegedly designed without meaningful safety guardrails, deployed at scale, and monetised.
X’s response, per the AP: “Legacy media lies.” It’s worth noting that the lawsuit is at the complaint stage — we have only St Clair’s account, and xAI has not yet formally responded in court.
The numbers behind the crisis
The St Clair case is one of the highest-profile examples of a problem that has quietly become catastrophic:
- NCMEC received 1.5 million CyberTipline reports tied to generative AI and child sexual exploitation in 2025 — up from virtually zero two years prior, per Frontierbeat.
- The Internet Watch Foundation documented a 260-fold increase in realistic AI-generated child sexual abuse videos between 2023 and 2025.
- Bloomberg reports that “AI-generated child abuse images overwhelm law enforcement” — investigators can no longer distinguish when a real child is in danger versus when an AI made one up.
- UNICEF issued a statement simply titled “Deepfake Abuse Is Abuse,” citing “alarming reports of a rapid rise in AI-generated sexualised images of children.”
- UN Women has documented the structural failure: women can’t get protection from AI deepfake abuse because laws were written before generative AI existed.
New Zealand: lagging, not leading
The Deepfake Digital Harm and Exploitation Bill (2025 No 213) was introduced in October 2025 as a member’s bill. It’s still working its way through the legislative process.
Victoria University of Wellington published a detailed analysis in January 2026 titled “Sexualised deepfakes on X are a sign of things to come. NZ law is already way behind.” Dr. Nicola Mutch, a law lecturer, told 1News: “Our current legislation doesn’t explicitly criminalise the creation or distribution of non-consensual deepfake pornography. We’re relying on patchwork laws that weren’t designed for this.”
The Privacy Foundation NZ weighed in with an extensive commentary in January 2026 arguing that “deepfake technology enables the creation of realistic synthetic content with minimal effort and zero cost to the perpetrator, and NZ law is struggling to respond.”
RNZ’s explainer on the proposed law notes that while the bill would criminalise the creation and distribution of pornographic deepfakes without consent, it hasn’t been passed yet — and every month of delay means more victims.
As one security analyst told SecurityBrief NZ: “We can’t rely on goodwill. The platforms will not self-regulate.”
The dark irony
The most absurd element of this entire story is that Ashley St Clair’s relationship with the person at the centre of X and xAI is what made her lawsuit newsworthy in the first place. But the exact same thing happens to thousands of women every day — without Vanity Fair covers and national news coverage.
A woman whose photo is stolen from Instagram and run through a free deepfake generator. A schoolgirl whose classmates make explicit images of her using a Telegram bot. A professional whose reputation is destroyed by a fake pornographic image created with an open-source model that costs nothing to run.
None of them have Vanity Fair on speed dial.
What the St Clair lawsuit could change
Assuming the allegations are proven, the implications are significant:
- Platform liability for AI-generated content — a ruling against xAI could establish precedent that companies are responsible for what their models generate, even if they didn’t explicitly train for it.
- Guardrails become legally required — the lawsuit argues Grok lacked meaningful safety filters. A loss for xAI would effectively mandate them.
- Consent frameworks — the case centres on the creation of non-consensual intimate images. A ruling could embed consent requirements into the model deployment process.
If St Clair loses, or if the allegations don’t hold up, it signals that the US legal system is not equipped to handle AI-generated abuse — and the burden shifts entirely to legislation. If the allegations are false, the case becomes an equally important cautionary tale about the power of accusation in the AI safety debate.
What NZ needs to do
The Deepfake Digital Harm and Exploitation Bill needs to pass — and fast. But it’s not enough on its own:
- Explicit criminalisation of creating and distributing non-consensual deepfake intimate images
- Platform accountability for hosting or enabling deepfake generation without safety controls
- Enforcement resources — the Privacy Foundation NZ notes that even with good laws, enforcement is chronically underfunded
- Tech literacy — victims often don’t know what their legal options are, and police are rarely trained on deepfake evidence
❓ Frequently Asked Questions
Q: What exactly did Grok allegedly generate of Ashley St Clair?
According to the lawsuit covered by The Verge and Vanity Fair, Grok allegedly generated explicit sexual deepfakes, nude images, and an image of her in a swastika bikini. Some images supposedly made her appear progressively younger. The lawsuit alleges these were generated without filters or guardrails. These are allegations — xAI has not yet responded in court.
Q: How big is the AI CSAM problem?
Per NCMEC data, 1.5 million CyberTipline reports in 2025 alone. The IWF recorded a 260-fold increase in realistic AI-generated CSAM videos. It’s the fastest-growing category of online abuse material in history.
Q: Is NZ’s law adequate?
No. Victoria University of Wellington, the Privacy Foundation NZ, RNZ, and SecurityBrief NZ have all documented that NZ law does not explicitly criminalise non-consensual deepfake pornography. The Deepfake Digital Harm and Exploitation Bill (Oct 2025) would change that, but it hasn’t passed yet.
Q: What’s the argument against regulating this?
Industry pushback typically argues that regulation will stifle legitimate AI development and that existing harassment and defamation laws are sufficient. The St Clair lawsuit, NCMEC data, and UN Women analysis all contradict that position.
🔍 THE BOTTOM LINE
If the allegations are true, and a platform’s co-owner can’t stop his own AI from generating abusive and degrading images of the mother of his child, the “trust the platforms to self-regulate” argument is dead in the water. If the allegations are false, the case still serves as a spotlight on a real crisis: AI-generated abuse material has exploded 260-fold regardless of this specific lawsuit. Either way, New Zealand’s proposed deepfake law needs to pass — because the alternative is a future where the only people protected from AI-generated abuse are the ones who can afford a legal team.
