HK Privacy Watchdog Joins 60-Nation Coalition to Combat AI Deepfake Violence

The Hong Kong Office of the Privacy Commissioner for Personal Data (PCPD) has signaled a significant shift in its regulatory posture by joining an international coalition of over 60 organizations to confront the rising tide of technology-facilitated violence (TFV). This move underscores a growing global consensus that existing regulatory frameworks are ill-equipped to handle the speed and sophistication of generative artificial intelligence. By co-signing this statement in late February 2026, the PCPD is moving beyond localized enforcement toward a unified global standard for AI safety, specifically targeting the disproportionate impact of deepfakes on women and girls.

The scale of the crisis is defined by a stark data point: approximately 90% of non-consensual deepfake pornography currently circulating online depicts women. This is not merely a new category of digital crime but an evolution of harassment that has transitioned from telephone-based abuse to cyberstalking, and now to AI-generated content. The core of the current regulatory argument, as championed by figures like Clarissa Lui, is that AI has not created these patterns of abuse but has 'supercharged' them, lowering the barrier to entry for malicious actors while increasing the potential for viral, permanent harm.

“

The scale of the crisis is defined by a stark data point: approximately 90% of non-consensual deepfake pornography currently circulating online depicts women.

From a RegTech and compliance perspective, the focus is shifting from reactive 'notice and takedown' models to a proactive 'safety by design' philosophy. Current platform governance is frequently criticized for being opaque and retraumatizing for victims. Many major tech platforms, including those hosting advanced AI models like Grok, maintain safety policies that often fail in practice due to delayed responses or ineffective reporting pathways. For legal professionals and tech developers, this implies a future where regulatory compliance will require embedding gender-centric safety protocols directly into the architecture of AI models during the training and design phases, rather than as an afterthought.