Measures against Perpetrators who Adopt Deepfake Images or Videos to Commit Offences

Ms Hany Soh asked the Minister for Home Affairs whether there are sufficient protective measures against perpetrators who appear to be increasingly adopting deepfake images or videos to commit various offences, including making of threats and extortion.

The Ministry of Home Affairs has been working with the Ministry of Communications and Information, the Infocomm Media Development Authority and the Cyber Security Agency of Singapore to address crimes which involve deepfakes, or which are otherwise facilitated by Artificial Intelligence.

The Online Criminal Harms Act (OCHA), allows the Government to issue directions to online service providers to prevent online criminal activities from interacting with or reaching Singapore users. These include deepfake images or videos used to commit offences, such as the making of threats and extortion.

OCHA also allows us to require providers of designated online services to put in place measures to disrupt scams and malicious cyber activities, including those facilitated by deepfakes. These measures will be stipulated in the Codes of Practice which will be introduced this year.

In addition, the Protection from Online Falsehoods and Manipulation Act (POFMA), enables the Government to issue corrections and label deepfakes that are deceptive or misleading to the public, where there is public interest to do so. POFMA also allows the Government to issue disabling directions to restrict access by Singapore end-users to such deepfakes.

The Government will also continue to strengthen our technical capabilities to deal with these threats. For example, the Singapore Police Force is working with the Home Team Science and Technology Agency to develop technologies to detect AI-generated audios and videos. The Centre for Advanced Technologies in Online Safety, an initiative by the Ministry of Communications and Information and hosted by the Agency for Science, Technology and Research, is also building tools to detect harmful online content, including deepfakes. That said, this is a serious challenge and technologies will continuously evolve. It is difficult to pick up all cases.