The rule of law for AI in elections

It is crucial to have strong legal frameworks and ethical guidelines for AI use. — Others

LAWS to govern the use of artificial intelligence (AI) in electoral campaigns should be in place due to its potential of being adversely implemented, says the Bar Council.

Bar Council’s technology, cyber and privacy law committee chairman Sarah Yong Li Hsien says while the Election Offences Act 1959 stipulates various offences which criminalise corrupt practices and acts that would impinge on a person’s free will when it comes to elections, more advanced methods can be used through AI to carry out such offences.

Malaysia’s situation is unlike in Europe, where the risk of AI is recognised through the drafting of the AI Act last month, which seeks to significantly bolster regulations on the development and use of AI, according to a report by the World Economic Forum.

“Yes, there should be laws to govern the use of AI for election purposes. With AI, there are now advanced methods that could deceive and manipulate masses on a very personal and direct manner, which in effect impinges their free will in voting.”

She says among the possible abuse of AI includes the employment of bots to impersonate real individuals and profiling target groups based on sensitive personal data.

“Some methods [that can be used] are AI-generated photos and videos, AI bots personating as people on social media, targeted political advertising and profiling based on personal data, and even sensitive information such as race, religious beliefs and political views.

“Our current laws are arguably restrictive, have yet to catch up with the times, and may not be wide enough to cover these abuses due to advancements in technology, especially AI.”

The Science, Technology and Innovation Ministry has said that it is looking into the possibility of regulating AI in Malaysia, which may also encompass elections.

Minister Chang Lih Kang says there is a need for a clear legal framework to regulate the possible abuse of AI in elections through slanderous content or misinformation.

“It is crucial to have strong legal frameworks and ethical guidelines for AI use. This could include laws that mandate transparency about the source of information and severe penalties for those who use AI tools to spread false information,” he says.

Yong says while Malaysia currently does not have a specific law to regulate AI, the Communications and Multimedia Act 2010 (CMA) and the Personal Data Protection Act 2010 (PDPA) may work to address any outcome from its potential abuse.

“There is no standalone law governing the use of AI. How-ever, an adverse outcome from potential abuse of AI can arguably fall under the ambit of Section 211 of the CMA, which prohibits indecent, obscene, false, menacing or offensive content, as well as Section 233 of the same Act that prohibits the making, creation, solicitation and transmission of communications that are obscene, indecent, false, menacing or offensive.”

As for the PDPA, its jurisdiction may not be enough to address the misuse of personal data in elections.

“We also have the PDPA that makes it illegal for personal data to be misused commercially in circumstances such as if consent was not given by the data subject for their data to be used in a certain way or for a certain purpose.

“Nevertheless, the scope of the PDPA is restricted to commercial transactions and elections may arguably be outside its scope,” Yong says.