Singapore has proposed new laws to outlaw the usage of digitally manipulated content material, particularly deepfakes, throughout elections.
The brand new authorized steps will safeguard “the integrity and truthfulness of illustration” during elections, stated the Ministry of Digital Growth and Data (MDDI) in a statement on Monday. The proposed invoice was launched in parliament and will probably be tabled for a second studying on the subsequent sitting.
It goals to supply new laws to safeguard the native inhabitants in opposition to digitally manipulated content material, together with audio, video, and picture generated by artificial intelligence (AI), in any other case often called deepfakes.
Additionally: Businesses still ready to invest in Gen AI, with risk management a top priority
Developments in generative AI (Gen AI) know-how have opened up alternatives, but in addition current information risks, with risk actors leveraging it to create and unfold misinformation, MDDI stated. It pointed to a rise in malicious deepfakes utilized in scams and extortion in Singapore.
Whereas online falsehoods can be dealt with by the nation’s present Safety from On-line Falsehoods and Manipulation Act, “focused levers” are essential below its On-line Election Promoting Act to handle deepfakes that misrepresent candidates, the ministry stated. The latter laws defines any content material accessible on-line that may be “moderately” deemed to advertise or improve the standing of political events or candidates or prejudice the electoral prospects of different political events or candidates.
If left unaddressed, deepfakes can compromise the integrity of Singapore’s electoral course of, MDDI stated. It added that voters ought to have the ability to make informed decisions based on facts, not misinformation.
Additionally: Artificial intelligence, real anxiety: Why we can’t stop worrying and love AI
The proposed legislative measures will prohibit publishing digitally generated or manipulated on-line election promoting that “realistically depicts” candidates saying or doing one thing they didn’t say or do. This consists of content material that makes use of each AI and non-AI strategies, similar to modifying through picture modifying, dubbing, or splicing.
The brand new measures will apply solely to on-line election promoting that options people working as candidates within the election.
The proposed laws will enable corrective instructions to be issued to people who publish such content material in addition to social media platforms, to take down offending content material. Directives will also be despatched to web service suppliers to disable entry by Singapore customers to such content material throughout the election interval.
Failure to adjust to corrective instructions has been deemed an offense and is punishable by a positive, imprisonment, or each if convicted.
The invoice additionally will enable candidates to submit requests for content material to be assessed for potential breach and corrective instructions issued accordingly. Candidates who’ve been misrepresented by such content material could make a declaration to attest to the veracity of their declare, MDDI stated.
It will likely be unlawful for candidates to knowingly make false or deceptive declarations of their request on the content material in query, the ministry stated. It added {that a} positive or different penalties could also be meted out ought to candidates be discovered to have dedicated such practices.
Additionally: Can governments turn AI safety talk into action?
The Singapore authorities additionally plans to introduce a code of follow mandating particular social media suppliers to implement measures to forestall and fight the usage of digitally manipulated content material on their platforms.
Particulars for this will probably be established within the coming months, with discussions to happen between the regulator and social media operators, MDDI stated.
“Past elections, deepfakes and different types of manipulated content material can pose critical hurt to on-line customers,” it stated. “Creators and distributors of AI-generated content material have a accountability to foster a secure data house, each throughout and outdoors of elections.”
Additionally: How Singapore is creating more inclusive AI
Singapore is anticipated to carry its normal elections early subsequent 12 months or earlier than November 2025.
South Korea adopted a 90-day ban on the usage of deepfakes in political marketing campaign content material for its April 2024 legislative election.