Generative AI, deepfakes will ‘pollute’ election campaigns, CSE warns

Canada’s cybersecurity watchdog is warning fake images and videos created by artificial intelligence (AI) will “very likely” be used to try to undermine voters’ faith in democracy in upcoming election campaigns.

In a new report published Wednesday, the Communications Security Establishment (CSE) stated AI-created deepfakes – computer-generated images and videos that show events that did not take place – “will almost certainly become more difficult to detect, making it harder for Canadians to trust online information about politicians or elections.”

“Despite the potential creative benefits of generative AI, its ability to pollute the information ecosystem with disinformation threatens democratic processes worldwide,” the agency wrote.

“So to be clear, we assess the cyber threat activity is more likely to happen during Canada’s next federal election than it was in the past,” CSE chief Caroline Xavier said.

The report also warned that cyberattacks are increasing around the world, that Russia and China are responsible for most attributed attacks and that the two countries have accessed voters’ personal information around the world.

And it says pro-Russia state-affiliated cyber actors have targeted elections in countries that provided assistance to Ukraine and specifically says the Kremlin could target the upcoming 2023 and 2024 European elections.

It says Canada is also very much at risk because Ottawa supports Ukraine, because Canada is a member of NATO and other prominent global groups and because Canadians are among the most internet-connected voters in the world.

Xavier said increased tensions between Canada and a hostile state leading up to a federal election would likely lead to cyber threat activity – including an online influence campaign.

And Xavier told reporters that foreign adversaries are increasingly using AI to create and spread disinformation.

“AI-generated deepfakes can create a video of something that never happened and put words into politicians’ mouths that they never said,” Xavier said, speaking at a press conference in Ottawa.

She added that social botnets (networks of accounts that spread items on social media) “enhanced by AI can artificially amplify disinformation that’s already out there to fan the flames of political discord and push people to more extreme positions.”

While the technology isn’t widely used yet, the report states the CSE expects use to increase in the next two years.

Xavier said attempted online interference has become “ubiquitous” to elections.

The CSE report examined 146 elections since 2015 and concluded that 26 per cent suffered cyber interference.

Rajiv Gupta, associate head of the Canadian Centre for Cyber Security, said no Canadian elections were impacted by “cyber-enabled incidents.”

Xavier said hack and leak operations, when actors gain access to confidential material and post it publicly, are seven times more likely than operations aimed at compromising voting processes.

Russia and China were the only identified states involved in those efforts, she told reporters.

But the two countries only account for 15 per cent of noted cyber activities – leaving 85 per cent unattributed, and the report says many actors are increasingly using third-party actors for misinformation campaigns.

“Foreign adversaries are getting better at hiding their identities,” Xavier said.

“They increasingly outsource their activities to avoid attribution.”

Despite that, Xavier said CSE had the capability and the legal ability to “conduct defensive cyber operations.”

“We are wide awake to threats,” she said.

Defence Minister Bill Blair Canada “has a robust system to detect, to identify and to communicate when misinformation is presented to Canadians.”

Blair said the report helps Canadians know that “perhaps what they’re seeing on social media isn’t entirely trustworthy and they should dig deeper to get the facts.”

“Cyber threat activity poses a real and growing threat to Canada’s democratic processes,” the report said.

In a statement, Elections Canada told Global News the Canada Elections Act does not specifically refer to deepfakes. It said the Act does prohibit fake images and videos in certain circumstances — but only if the images and videos are purportedly made by any Elections Canada worker, political parties and people associated with them, candidates or prospective candidates.

Elections Canada also said it leverages “the extensive monitoring capabilities” that are supported by the Canadian Centre for Cyber Security, “which allows for anomalies to be quickly detected across all platforms and systems.”