HK privacy watchdog raises alarm over Grok's potential for harmful content

Hong Kong's privacy watchdog has raised concerns that the artificial intelligence (AI) chatbot Grok can be used to generate indecent or malicious photos and videos, reminding citizens to use AI safely.

In a statement on its website, the Office of the Privacy Commissioner for Personal Data (PCPD) said the misuse of AI has raised concerns in various jurisdictions. The PCPD is contacting the relevant organisations to understand the situation.

Don't miss: MCMC seeks meeting with X as it prepares legal action over Grok

This follows a surge in a viral “remove clothes” trend on X, formerly Twitter, in which users have uploaded photographs of real people - including women and children - and asked Grok to generate images showing them partially or fully undressed. 

Malaysia and Indonesia have blocked access to Grok over concerns about non-consensual sexual deepfake content. Meanwhile, the UK's media regulator announced an investigation into X for allegedly generating sexually explicit deepfake images in violation of its duty to protect the public from illegal content.

The PCPD has reminded the public that when providing personal data to AI chatbots for generating content, they must comply with the Personal Data (Privacy) Ordinance (PDPO) and relevant data protection principles. Improper or malicious use of AI chatbots to create indecent or harmful photos or videos may violate PDPO requirements and could constitute other criminal offenses.

To safeguard personal data privacy, citizens are advised not to share their own or others' personal data with AI chatbots arbitrarily. They should obtain voluntary consent from others before using their personal data to generate AI content and must never use AI chatbots to create illegal content, such as child pornography.

Furthermore, citizens should read the privacy policy and terms of use of AI chatbots to understand how personal data will be collected, stored, used, and shared. They should also be cautious about using content generated by AI chatbots, remaining aware of potential illegal, inaccurate, copyright-infringing, biased, or discriminatory content.

The PCPD has earlier published a leaflet entitled “10 tips for users of AI chatbots” and a guidance entitled “Abuse of AI deepfakes: Toolkit for schools and parents”, to help users protect their own and others’ personal data privacy and to use AI chatbots more safely.

Back in June, PCPD rolled out a series of promotional and educational activities to enhance the awareness of the public and organisations to AI security, including the running of thematic trams. Themed “AI security matters for all", the tram design prominently featured a lock symbolising AI security, alongside various icons relating to personal data. It also included information on the “AI security” hotline and a QR code linking to the “AI security” thematic website, thereby enabling members of the public to easily access relevant information.

Related articles:

MCMC seeks meeting with X as it prepares legal action over Grok
Indonesia officially blocks Grok as AI deepfake scrutiny intensifies across markets
Malaysia blocks Grok AI over sexual content and misuse