Political deepfakes are AI-generated digital content that depict a politician doing or saying something they did not do or say. They can also depict someone falsely expressing a view about a politician or political issue. Political deepfakes can be deceptive, sexually explicit or satirical. They can also draw positive and negative associations through imagery, analogies and associations. 

There are international concerns about the proliferation of deepfakes that are designed to mislead voters around elections. ​In March 2026, New South Wales (NSW) became the second Australian jurisdiction to enact legislative reforms. The Electoral Legislation Amendment (Elections) Act 2026 creates two new offences that prohibit​ a person from publishing or distributing a deepfake during an election period. 

This paper outlines the potential adverse impacts of political deepfakes, summarises the NSW reforms, and examines legislative developments in other jurisdictions. The paper also discusses other measures that can be used to combat political deepfakes.

Key findings

• Since 2017 one database has captured 2,201 global incidents of political deepfakes. Another database recorded 453 global incidents of political deepfakes in 2025.
• There was debate in parliament as to whether the new offences are consistent with the implied freedom of political communication in the Australian Constitution. 
• Other jurisdictions have also enacted political deepfake laws. These include South Australia, many states in the United States, and one Canadian province. 
• Other measures for combating political deepfakes include the use of watermarks, deepfake detection technology, removal of deepfakes by social media platforms, and public education and awareness.