Singapore proposes law to ban deepfakes during elections

Singapore has proposed new legislation aimed at curbing the use of deepfakes and other digitally manipulated content during elections, following concerns about the misuse of artificial intelligence (AI) to spread misinformation. The Elections (Integrity of Online Advertising) (Amendment) Bill, introduced by the Ministry of Digital Development and Information (MDDI) on Monday (9 September), would target AI-generated content that depicts election candidates saying or doing things they did not say or do. This includes manipulated content using non-AI techniques like Photoshop, dubbing, and splicing.

Safeguarding Elections from Misinformation

The Bill comes amid rising concerns globally over deepfake technology being used maliciously, including for scams and extortion. With Singapore's next General Election due by November 2025, the new legislation is designed to prevent misinformation from influencing voters’ decisions. A spokesperson for MDDI emphasized the importance of maintaining the integrity of election advertising during the sensitive period between the issuance of the writ of election and the close of polling. Under the proposed law, it would be a criminal offence to publish, share, or repost digitally altered content that meets three specific criteria:

1. The content is manipulated using AI or non-AI techniques.
2. The content is intended as online election advertising.
3. The content realistically depicts a candidate saying or doing something they did not.

Penalties and Enforcement

Should the Bill pass, violators could face fines of up to S$1 million for social media services that fail to comply with take-down requests from the Returning Officer, and fines of up to S$1,000 or imprisonment for individuals who do not comply. The Returning Officer would also have the authority to issue corrective directions to remove or disable access to offending content during the election period. Additionally, election candidates could request the Returning Officer to assess questionable content, although those who make false declarations could face penalties, including losing their seat.

Balancing Expression and Integrity

The Bill aims to strike a balance between freedom of expression and safeguarding the accuracy of election-related content. Minor digital modifications, such as beauty filters, and entertainment content like memes and animated characters, would not be included under the proposed law. Private communication, such as one-on-one chats or family group conversations, will not fall under the law, but larger, more public forums, such as large WhatsApp groups, may be subject to the prohibition based on the nature of the relationships within the group.

Deepfake Threats and Wider Context

MDDI highlighted previous incidents in Singapore where political office holders were targeted with deepfakes and digitally altered images. Earlier in 2024, Senior Minister Lee Hsien Loong warned of deepfake videos impersonating him in discussions on international relations. While existing laws such as the Protection from Online Falsehoods and Manipulation Act (POFMA) address general falsehoods, the new Bill is seen as a necessary addition to address the specific threat of deepfakes and AI-generated content during elections. Beyond the immediate election framework, the Infocomm Media Development Authority (IMDA) is working to introduce a legally binding code of practice for social media services to prevent and counter the abuse of digitally manipulated content.