The act of creating and/or sharing false, misleading, or sensationalized information is far from novel, with instances dating back to as early as the 15th centuryFootnote 1Footnote 2Footnote 3
Disinformation is false information that is deliberately intended to misleadFootnote 4Footnote 5
Deepfakes are media manipulations that are based on advanced artificial intelligence (AI), where images, voices, videos or text are digitally altered or fully generated by AIFootnote 6Footnote 7Footnote 8Footnote 9
While deepfake technology is being used to create wholesome, entertaining, and satirical content, governments need to consider the potential harms and/or threats to public safety that this technology poses. As this paper demonstrates, deepfakes warrant the attention, and action of democratic governments and those who value the freedoms and safety of life in a democratic country.
Advancements in AI
The Treasury Board of Canada (TBS) defines AI as information technology that performs tasks that would ordinarily require biological brainpower to accomplish, such as making sense of spoken language, learning behaviours, or solving problemsFootnote 10Footnote 11Footnote 12
In the last few years, AI has advanced significantly in its performance of these so-called “human tasks”. ChatGPT (openai.com) is a prime example of such an advancementFootnote 13Footnote 14
Text-to-image generation is another significant advancement for synthetic media, wherein an AI model creates a unique image based on user-inputted key wordsFootnote 15
AI has also advanced the speed at which human tasks are accomplished generally. AI-based program AlphaFold, for example, is assessed to have correctly predicted the structures of over 200 million proteinsFootnote 16Footnote 17
The Threat to a Canadian Future
Advancements in AI are rapidly improving the realism of deepfakes, as well as making them more difficult to detect and disseminateFootnote 18Footnote 19Footnote 20
Taking these factors into account, the question of whether deepfakes are an issue of concern for Canadians is raised. In response, one could consider a statement made by Nobel Laureate Maria Ressa:
Without facts, you can’t have truth. Without truth, you can’t have trust. Without trust, we have no shared reality, no democracy, and it becomes impossible to deal with our world’s existential problemsFootnote 21Footnote 22
Ultimately, the issue centers on facts. If a democratic society is unable to differentiate fact from fiction, then how is it going to survive? How is Canada going to function if there are different sets of unverifiable facts that different segments of the population believe in? If disinformation is unmanageable and/or unidentifiable, how is Canada going to develop solutions for those real problems? What does this mean for Canadian values and/or the Canadian way of life? Moreover, what happens when deepfakes are used for malicious purposes or with the intent to harm Canadians or their allies?
Harms Caused Using Deepfakes
The harms caused by deepfakes are illustrated by recent examples. In January 2023, a young woman named Blaire, a Twitch streamer and YouTuber better known as ‘QTCinderella’, discovered that a deepfake porn site was using her face/likeness in porn videos, along with the faces of other female Twitch streamersFootnote 23Footnote 24Footnote 25
In 2019, Rana Ayyub, an investigative journalist for the Washington Post, spoke out against a political party in India that was protecting the rapist of an eight-year-old Indian girlFootnote 26Footnote 27
Such examples of deepfake porn are not uncommon. Over 90 per cent of deepfakes available online are non-consensual pornographic clips of women; as of October 2022 there were over 57 million hits for ‘deepfake porn’ on Google aloneFootnote 28Footnote 29Footnote 30
Notably, not all harm caused by deepfakes is of the pornographic variety. For example, criminals have repeatedly used deepfakes of Elon Musk in fraudulent cryptocurrency giveaways, resulting in financial losses totaling in the millions of dollarsFootnote 31Footnote 32Footnote 33Footnote 34Footnote 35
Following the COVID-19 pandemic, the world has adapted to using virtual platforms to host meetings, interviews, classes, etcFootnote 36Footnote 37Footnote 38
Other AI Considerations
It is clear that AI is a powerful tool that can advance solutions and facilitate positive outcomes to problems. However, it can also equip an entity with the power to cause significant damage. Consider the following:
Privacy Violations. AI systems can collect, process, analyze, and store significant volumes of data. If a system is hacked or a security breach occurs, this data (e.g., medical history, biographic data, and/or banking information) can easily be stolen, manipulated, and/or extorted for nefarious purposesFootnote 39Footnote 40Footnote 41 Social Manipulation. AI can be used to track, analyze, and predict individuals’ online activities, which can make them vulnerable to manipulation or being compelled to engage in activities in which they would not otherwise participateFootnote 42Footnote 43 Bias. Humans, with their implicit and explicit biases, build, train, and test AI systems. These biases can be reflected in the decisions taken by AI systems, to the detriment of certain groupsFootnote 44Footnote 45 A Path Forward for Governments
Globally, an estimated 5 billion people use or have access to the internetFootnote 46Footnote 47
Deepfakes and other advanced AI technologies threaten democracy as certain actors seek to capitalize on uncertainty or perpetuate ‘facts’ based on synthetic and/or falsified information. This will be exacerbated further if governments are unable to ‘prove’ that their official content is real and factual.
Deepfakes and synthetic media can also facilitate psychological, reputational, and economic harmsFootnote 48Footnote 49
Indeed, a number of the foremost experts in AI—Yoshua Bengio, Elon Musk, and Geoffrey Hinton—have all emphasized the dangers that AI presents to the publicFootnote 50Footnote 51Footnote 52Footnote 53Footnote 54Footnote 55
Canada’s TBS has indicated that the Government of Canada is committed to using AI to support and/or improve some of the services it provides to Canadians in a manner that is compatible with the core principles of administrative law. This is the basis for TBS’ Directive on Automated Decision Making, which aims “to ensure that automated decision systems are deployed in a manner that reduces risks to clients, federal institutions and Canadian society, and leads to more efficient, accurate, consistent and interpretable decisions made pursuant to Canadian law”Footnote 56
Another positive step forward is the Department of Canadian Heritage’s Digital Citizen Initiative (DCI), “a multi-component strategy that aims to support democracy and social inclusion in Canada by building citizen resilience against online disinformation and building partnerships to support a healthy information ecosystem”. The DCI aims to help both Canadians and the Government of Canada to better understand disinformation, and its impacts in order to determine the appropriate actions to take and inform future policiesFootnote 57
AI capabilities will continue to advance and evolve; the realism of deepfakes/synthetic media is going to improve; and AI-generated content is going to become more prevalent. This means that governmental policies, directives, and initiatives (both present and future) will need to advance and evolve in equal measure alongside these technologies, including capacities to characterize and differentiate malicious AI-based content from prosocial and positive applications.
When it comes to drafting and implementing new policies, procedures and/or legislation, democratic governments are, perhaps by necessity, notoriously slow movingFootnote 58Footnote 59Footnote 60
