Deepfake Detection

State of play.

• Italy has moved from civil litigation to criminal enforcement on deepfake pornography. Following Meloni's 2024 lawsuit against two men who created deepfake pornography using her likeness, Italy has strengthened its AI laws to include prison terms for creators of harmful deepfakes — and Meloni's deliberate republication of a fabricated image of herself signals that awareness campaigns are now being treated as insufficient on their own (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Facial trademark registration is emerging as a proactive IP defense against AI-generated likeness abuse. Cole Palmer's successful photorealistic facial trademark registration with the UK IPO in November 2025 and Luke Littler's pending application signal that IP offices are opening the door to a new protective instrument for high-profile individuals .
• Litigation readiness for deepfake evidence is an open and unresolved question. Commentary from practitioners flags that courts, evidentiary standards, and authentication protocols have not kept pace with synthetic media capabilities .
• For counsel advising entertainment, sports, media, or public-figure clients, the practical baseline is that deepfake exposure now requires a layered response — criminal referral, civil enforcement, and proactive IP registration — because no single legal instrument is currently adequate.

Where things stand.

• Non-consensual AI-generated sexual imagery is disproportionately targeting women. Approximately 90 percent of non-consensual AI-generated sexual imagery depicts women; X's Grok tool generated an estimated 3 million sexualized images between December 2025 and January 2026, illustrating the scale of platform-enabled harm (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Italy is the most advanced jurisdiction on criminal deepfake enforcement. Prison terms for creators of harmful deepfakes are now on the books; the Meloni incident is the highest-profile test of whether enforcement follows legislative intent (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Facial trademark registration is gaining traction in the UK and EU as an anti-deepfake instrument. The UK IPO has accepted photorealistic facial marks for Cole Palmer; the EU has granted marks to model Maartje Verhoef; and the EU Grand Board of Appeal is reviewing the Jan Smit case, which will set the standard for photorealistic facial marks across the bloc .
• The evidentiary framework for deepfakes in litigation is unsettled. Practitioners are actively debating authentication standards, chain-of-custody requirements for synthetic media, and whether existing evidence rules are adequate to handle deepfake exhibits or deepfake-based attacks on authentic evidence .
• Experts are converging on cryptographic hardware authentication — not awareness or platform self-regulation — as the structural solution. The Meloni incident reinforced practitioner commentary that platform safeguards and education campaigns are insufficient; cryptographic provenance and aggressive legal enforcement are the proposed baseline (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).

Latest developments.

• Italian Prime Minister Meloni deliberately republished an AI-generated fabricated image of herself as a public warning, following Italy's enactment of prison terms for harmful deepfake creators and her own 2024 civil lawsuit against two men who created deepfake pornography using her likeness (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Luke Littler filed a UK IPO application to register his face as a trademark across categories including computer games and video games, following Cole Palmer's successful photorealistic facial trademark registration in November 2025 .
• The EU Grand Board of Appeal is reviewing the Jan Smit case, which will set binding standards for photorealistic facial marks across EU member states .
• Practitioner commentary in Above the Law flags that courts and evidentiary frameworks are not prepared for deepfake evidence challenges in litigation .

Active questions and open splits.

• Whether facial trademark registration can function as an effective anti-deepfake instrument. The UK IPO has accepted photorealistic marks, but the distinctiveness standard remains unsettled; the EU Grand Board of Appeal's Jan Smit ruling will be the first binding guidance on whether photorealistic likenesses qualify .
• Whether platform liability for deepfake-generation tools is actionable. X's Grok generating an estimated 3 million sexualized images raises the question of whether enabling tools — not just individual creators — can be reached under existing or emerging law (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Whether Italy's criminal enforcement model will be adopted by other jurisdictions. Italy has prison terms on the books; whether the EU AI Act, UK Online Safety Act, or US state statutes converge on a similar criminal standard is unresolved (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Whether existing evidentiary rules are adequate to authenticate or exclude deepfake evidence. No settled standard exists for deepfake authentication in litigation; courts have not yet produced a durable framework for chain-of-custody or expert-witness requirements for synthetic media .
• Whether cryptographic hardware authentication will become a legal or regulatory requirement. Experts advocate it as the structural fix, but no jurisdiction has yet mandated it; the gap between expert consensus and regulatory action is wide (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).

What to watch.

• The EU Grand Board of Appeal's ruling in the Jan Smit facial trademark case — the outcome will set the standard for photorealistic facial marks across the EU and will directly affect the viability of facial trademark as an anti-deepfake strategy for clients.
• The outcome of Luke Littler's UK IPO facial trademark application and whether the IPO extends the Cole Palmer precedent to additional categories .
• Whether Italy's criminal enforcement of its deepfake prison-term statute produces the first conviction, and whether other EU member states move to harmonize on a criminal standard (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Whether any jurisdiction moves to impose platform liability for AI tools that generate non-consensual synthetic imagery at scale, following scrutiny of X's Grok (→ Meloni Posts AI-Generated Nude to Warn of Deepfake Danger).
• Whether courts in the US, UK, or EU produce the first ruling on deepfake evidence authentication standards in civil or criminal litigation .