Canada Tables Bill to Criminalize Election Deepfakes
Canada's federal government has tabled Bill C-63, the Online Harms Act, which would criminalize the spread of deepfakes and other synthetic media intended to influence elections. The legislation proposes amendments to the Criminal Code and the Canadian Human Rights Act. This move signals a growing trend of national governments enacting specific laws to protect election integrity from AI-driven disinformation.
- The bill proposes a new Digital Safety Commission to enforce the rules, with the power to levy penalties up to 6% of a company's global revenue or C$10 million, whichever is greater. A Digital Safety Ombudsperson will also be established to advocate for users. - Introduced by Justice Minister Arif Virani, the legislation is a revised version of a previous bill, C-36, which failed to pass before the 2021 election. The current bill is modeled after the UK's Online Safety Act 2023 and follows extensive consultations. - The Act would require social media services, including adult content and live-streaming platforms like PornHub and Twitch, to remove specific harmful content within 24 hours of it being flagged. This includes non-consensual intimate content, which explicitly covers deepfakes of a sexual nature. - Beyond election-related content, the bill targets seven categories of harm, including content that sexually victimizes a child, induces self-harm, is used for bullying, or foments hatred. - The legislation includes controversial amendments to the Criminal Code, creating a new standalone hate crime offense and increasing the maximum punishment for advocating genocide to life imprisonment. - Critics, including the Conservative Party and civil liberties groups, have raised concerns about potential overreach, censorship of lawful content, and the severity of proposed penalties, such as life imprisonment for certain hate-motivated offenses. - Compared to the EU's AI Act, which is broader in scope, Canada's proposed legislation is more specifically focused on election disinformation. Canada's other AI-related bill, the Artificial Intelligence and Data Act (AIDA), takes a risk-based approach similar to the EU's but is considered more principles-based and less prescriptive.
