May 1, 2019

How fake is your news? Safeguarding democracy in the digital age

The role of mainstream media as providers of unbiased information is fading fast as people choose to get their information from sources that reflect their pre-existing views. At the same time, it's becoming easier to manipulate large masses of people through digital technologies. How will these combined forces play themselves out in the voting booth?

As they watch the bitter fracturing of American political consensus over what is true versus fake in the era of social media, Canadian voters may well wonder what’s in store for their own democracy.

“A lot of our fundamentals in terms of the administration of elections are very strong in Canada, so I think there’s plenty in our favour,” says Dr. Lisa Young, PhD, a political science professor in UCalgary's Faculty of Arts and member of the School of Public Policy. “But if someone had told me in 2015 that I would be as worried about American democracy as I am right now, I wouldn’t have believed them.”

Nearly three years after the election of Donald Trump as president of the U.S. in 2016, Canada is slated to hold its 43rd federal election by Oct. 21, and Albertans elected a new provincial government on April 16.

As stable as Canadian democracy has seemed, digital communication technologies are entering a new era, says Young. They're not only testing the limits of what constitutes free speech, they're giving everyone from average people to foreign governments the ability to deliberately mislead or misdirect voters in increasingly powerful ways, she says.

Deepfake: putting words in peoples' mouths

For example, using artificial intelligence techniques such as machine learning and neural networks, researchers at the University of Washington created a digital prototype in 2017 that synthesized a fake video of former U.S. president Barack Obama.

“They put in words from pieces of different things that Obama really said in different speeches, and they used them to train the prototype to create a video that is all fake, but looks very realistic,” says Dr. Hadi Hemmati, PhD, an assistant professor at the Schulich School of Engineering who researches how to apply machine learning and data science to make software more secure and reliable. “The voice sounds exactly as Obama’s, but the video is all synthesized and he never said those things.”

A man with a TV for a head

Deepfake video technology can be extremely deceiving.

Such “deepfake” video technology has already been used to put the faces of celebrities, and even average people, on the bodies of porn actors. Fearing misuse, a non-profit company partly backed by Elon Musk recently refused to release its full research about an artificial intelligence model that could possibly be used to write plausible deepfake news stories.

The potential erosion of trust even in what has been traditionally regarded as incontrovertible evidence is “what worries me the most as a political scientist,” says Dr. Melanee Thomas, PhD, an associate professor of political science in the Faculty of Arts. “It's going to be difficult for us to collectively understand what the actual facts are in a particular situation,” she says.

Heavily biased sources

The ability of Canadians to responsibly exercise their right to vote is increasingly at risk of being swamped by rapid changes in digital communication technologies, says Young, who is returning to her role as a researcher after several years as Dean of the Faculty of Graduate Studies at UCalgary. “A lot of my previous work was on election finance laws, and my question now is whether the regulations we have in place are even meaningful in the digital age,” she says.

“It’s not just in terms of regulation, but in terms of understanding what is going on so that we can be educated voters in this rapidly changing landscape, and to understand the ability of political parties, data analytics firms and others to shape our view of reality.”

As 2018’s Cambridge Analytica scandal demonstrated, it's now possible to use technologies such as artificial intelligence and machine learning to analyze the personal online data of millions of people without their consent.

More voters are getting their political information from sources ranging from Twitter, Facebook and YouTube to alternative news websites, says Thomas. “If you can build a profile of somebody based on what they are doing or saying on the Internet, you can get a pretty good idea of who they are and how they might vote,” she says.

Not only can such sources be heavily biased to one point of view, they may try to deliberately mislead voters, she says. While the political right is often blamed for such behaviour, the left is not immune, says Young.

To what extent will Canadian parties engage in these kinds of tactics?

The New York Times recently reported that progressive Democrats created fake Facebook pages to divide Republican voters in Alabama. “If you don’t do it, you’re fighting with one hand tied behind your back,” one activist was quoted as saying. “You have a moral imperative to do this — to do whatever it takes.”

Young says it will be a “fascinating question” to see if such thinking catches on in Canada. “We have a federal election coming up this fall, so to what extent will Canadian parties engage in these kinds of tactics?” she says. “Will there be organizations that are supportive of parties, but not co-ordinated with them, that intervene in this kind of way using social media to spread fake news?”

The growth of such third-party groups and their use of social media has already caused concern in Canada. During last year’s Ontario provincial election, not only did the lower cost of advertising on Facebook allow them to legally work around rules meant for expensive traditional TV advertising, they denigrated political opponents in ways that Canadian politicians have traditionally avoided.

Even official parties have used Facebook’s ability to micro-target specific groups of voters so that individuals see different political ads with unique messages aimed at them, rarely viewing the potentially contradictory ads aimed by the party at other voters. “Everything is becoming fractured, so understanding political behaviour and putting it into context is becoming really challenging,” says Thomas.

The shrinking role of mainstream media

It was easier for voters to together assess the effects of political news and reach a consensus “when we were all watching the same six o’clock TV news, in part because everyone got the same set of facts,” says Thomas. But the central, unifying role once held by traditional mainstream media such as TV, newspapers and radio is fading, increasingly changing how voters behave, she says.

A politician such as Trump (or anyone else) can now send unfiltered political messages directly and instantaneously to millions of supporters on Twitter, bypassing the fact-checking and context placement of traditional media even as he dismisses such reporters as purveyors of fake news, she says.

At the same time, the ability of traditional media to competently cover politics has eroded as an ever-greater slice of advertising dollars goes online, forcing the layoff of staff as newsrooms downsize or close, says Thomas. Some analysts are predicting that in the next few years, most of Canada’s major cities will likely no longer have a hometown daily newspaper, with the resulting vacuum filled by social media platforms built on sharing rather than creating news reports.

In that act of sharing, social media can deepen political conflicts by creating “echo chambers” or bubbles of information around each user that insulates them from differing opinions, says Hemmati. Social media is built on algorithms that train a particular platform to automatically deliver content to each user based on their individual interactions with the platform, he says.

“You only see stuff that you subconsciously have set to see,” says Hemmati. “There's no misinformation — there is nothing fake in it — but you only see that particular information, and you think that is the entire news, but that's just part of it.”

Taking advantage of such bubbles with political messages honed through data analysis is at the heart of the Cambridge Analytica scandal. The now-defunct company used Facebook data to advise clients — including the Trump campaign, and the pro-Brexit group in favour of the United Kingdom leaving the European Union — on how to win over voters.

Election interference by foreign actors

It is possible the data and analysis also helped inform the Russian disinformation campaign interfering in the 2016 U.S. presidential election. U.S. intelligence officials have said that not only did Russia hack e-mail accounts connected to the Democratic Party, it also directed the dissemination of fake news through social media.

Facebook said about 126 million Americans may have viewed posts by Russian-based operatives. Many such social media posts may not even have been from a real human being, but from a bot, which is software able to run automated scripts or tasks over the Internet at a rate much faster than a person.

During the U.S. presidential election, about 50,000 Russian-linked bots sowed discord through Twitter. Canada’s Communications Security Establishment recently warned that foreign countries are “very likely” to try to sway Canadian public opinion with online misinformation in 2019.

Young wryly remembers what she and some colleagues regarded 19 years ago as a cutting-edge way for politicians to micro-target voters. “We observed that parties could purchase subscription lists from a printed, not an online, magazine and then mail a letter — you know, a physical letter,” she says, laughing.

“The letter would make an appeal to them based on what was assumed were their interests. You might actually have no interest in politics, but a party would know you were interested in, say, hunting because you subscribed to certain magazines, so you might be open to hearing an appeal against a gun law.”

Just like Facebook ads today, these letters were targeted to a voter’s presumed interests, and were private communications between parties and voters, with little opportunity for journalists or other parties to challenge their content, says Young.

Political parties gathering data, too

Before helping establish Cambridge Analytica in 2013, whistleblower Christopher Wylie worked for Canada’s federal Liberal party as a volunteer and researcher. Wylie’s contract was not renewed by the party, said a former insider, due to concerns over his advocacy of an early form of data harvesting.

But the Liberal Caucus Research Bureau said in a statement the party later paid $100,000 in 2016 to Wylie’s company, Eunoia Technologies, as part of a pilot project that partly involved setting up social media monitoring tools.

Other goals included organizing and designing national samples of Canadians to explore responses to government policy priorities and other issues of national importance. After preliminary work was completed, the bureau decided not to proceed further, adding no data was provided by the bureau to either Eunoia Technologies or Wylie, who is a Canadian.

But when it comes to keeping tabs on Canadian voters, Young says it is hard to say what cards our country’s federal and provincial political parties are actually holding. “The problem is we don’t entirely know what kind of information they’ve got, or how they’re using it,” she says.

The Canada Elections Act gives political parties and candidates access to voters lists, including names and addresses, which they use as the basis for databases. While Elections Canada also provides guidelines for how the lists can be used, and emphasizes the importance of safeguarding confidential information, political parties often supplement this basic information with data derived from other sources.

“Any sorts of interactions with the party are also likely to be captured in databases, and they can further extrapolate based on where you live the likelihood you support the party,” says Young. “They may be able to cross-reference with information from other organizations to try and have a sense of the sorts of things you are interested in, but it may also extend beyond that if they’ve cross-referenced it to material obtained through social media.”

Canada recently approved Bill C-76, the Elections Modernization Act, which requires each party to create and post on its website policies about how they will protect personal information. But they are exempted from following laws such as the federal Privacy Act or the Personal Information Protection and Electronic Documents Act.

“I am not assuming it is necessarily nefarious to try to get people to vote in a particular way, but I would say Canadians usually react really strongly to the idea that somebody is gathering information about them without their consent, or was watching their activities on the Internet to build a profile about them and then using it politically,” says Thomas.

B.C. is the only jurisdiction in Canada that has personal information legislation covering its political parties. “I think one of the issues in Canada has been that our lawmakers, the political parties at the national and provincial levels, are setting the rules themselves and are pretty content to exempt themselves from regulation,” says Young.

“They might be looking at regulating some of the digital platforms, such as social media, but they certainly don’t want to apply privacy legislation to their own internal practices, so that is a huge concern. It’s just a matter of time until a political party is hacked and Canadians’ personal information is exposed.”

Security vulnerabilities

Political parties are seen by intelligence experts as being particularly vulnerable to cyber attacks that could provide hackers with information that could be used to misdirect voters. Under the proposed Bill C-59, Canada would gain the ability to conduct defensive and offensive cyber measures against foreign hackers trying to interfere with Canadian elections.

Although much more voter information is legally available to politicians in the U.S., including at times which party each voter supports, the fact that such data is restricted in Canada likely makes it “one of these places where people will be keen to be getting that kind of information,” says Thomas.

We bank online — why wouldn’t we vote online?

As she reviews the current situation, Thomas remembers how she was once in favour of allowing Canadian citizens to vote online, rather than the existing practice of going in person to polling stations to physically mark a paper ballot under the supervision of electoral officials.

A ballot emerges from a mobile phone

One advantage of paper ballots is that they can't be hacked, say experts.

“Five years ago, I would have been like: ‘We bank online — why wouldn’t we vote online?’”, she says. “I thought it would be a simpler way to go. I also thought it would be a way to spark voter turnout in populations who typically wouldn’t vote or participate, such as people with a lower socioeconomic status who now have access to things like tablets and smart phones.

“But now, I’m more cautious about online voting, in part because a paper ballot can’t be hacked. It’s not something I would support at this point precisely because of the security reasons, and my view has changed considerably.”

Anti-interference legislation

The Elections Modernization Act aims to prevent attempts by either foreign or domestic “bad actors” to interfere in Canadian elections through things such as social media. It will require online platforms such as Google and Facebook to create a registry of digital ads placed by political or third parties, as well as ban advocacy groups from using funding from foreign entities to conduct partisan campaigns.

The federal government also recently announced a further series of measures aimed at protecting elections from foreign interference. Although these include creating a team of bureaucrats who will warn the public if a large-scale threat is discovered, the measures do not include imposing new regulations on social media platforms to prevent the spread of misinformation.

The Internet: a double-edged sword

The “beautiful” early promise of the Internet of the 1990s and 2000s — that its technologies would lead to a more informed, balanced and democratic world that would help level the playing field for average or disenfranchised people — has darkened, says Thomas.

“The hope was that it would increase the number of sources of information we had, so that a diligent citizen could read widely and come up with their own well-informed view about what is going on,” she says. “It turns out this is not what’s happening.”

Rather than being passive consumers of information, as they were while watching television or reading a newspaper, voters must now spend increasing amounts of time evaluating the legitimacy of online sources, along with seeking out and evaluating political information, says Thomas.

“We went from a thing like television, where the audience was super passive and everyone basically got the same set of facts, to the current format where it’s much more interactive, and as a result, we now have a much more significant role to play as individual citizens in how our political environment gets formed and what we do with it,” she says.

We don’t have very many skills anymore for civil debate or discourse . . . 

Along with the rest of the world, Canadians are not adapting to this new reality “as well as we could,” says Thomas. “I think it requires individuals to not only be much more critical and skeptical, but also kinder.

“In general we don’t have very many skills anymore for civil debate or discourse where people can fully disagree with each other on pretty fundamental things, but still have a reasonable conversation where they are open-minded enough to be able to change their minds about some things.

“People often get very hostile and upset right away when presented with some facts that might correct some views they might hold, and that actually causes them to retrench even further into their beliefs.”

Young says she “certainly didn’t imagine the kind of social media world that we now live in. I’m worried that things will become worse.”

But Hemmati sees the current digital age as a generally positive development, even as he recognizes its pitfalls. “In the past, when we didn’t have the Internet, your voice could not be heard outside your own community,” he says.

“All you could do was go to something like a political rally and try to get your voice heard by your local MP or MLA, but now, you can literally do anything. Your views can go viral and there is no stopping for that, and it is only possible because of the Internet and social media.

“It’s like what happened with the #MeToo and similar movements in recent years. They were all started by individuals, and there was no central power that supported them, so we don’t need the support of anything. Many good movements can happen just because of social media.”

Hemmati says improvements in software— such as creating algorithms that place other viewpoints within social media bubbles, and better methods of detecting fake information — can help combat the negative side of the digital age.

He sees it as a kind of digital arms race, with constant vigilance being required to create new countermeasures as new threats such as deepfake technologies arise. “There should also be some kind of regulation by the state for these companies, such as Facebook, Twitter, etc., to define the borderlines of legal versus illegal newsfeeding strategies,” he says.

Canada needs to figure out “exactly where the threats in the Canadian context are, and to what extent we need to adapt our legislation,” says Young.

“There is always a balancing act with free speech and not wanting to interfere with the ability of parties and candidates to get their message across during the election. The filter should just be stopping misleading information, and that’s the tricky part. I don’t know that we've really fully come to terms with this in Canada yet.”


–  –  –  –  –

Explore our programs

Political science

Engineering

Participate in a research study


–  –  –  –  –

ABOUT OUR EXPERTS

Dr. Lisa Young, PhD, is a political science professor in UCalgary's Faculty of Arts, and former dean of the Faculty of Graduate Studies. Her research focuses on political parties, election finances and women’s participation in political life. Read more about Lisa

Dr. Hadi Hemmati, PhD, is an assistant professor in UCalgary's Schulich School of Engineering. His research areas are in software engineering and data science. Read more about Hadi

Dr. Melanee Thomas, PhD, is an associate professor in the Department of Political Science in UCalgary's Faculty of Arts. Her research focuses on political behaviour, Canadian politics, women and politics, electoral behaviour, and research methods. Read more about Melanee

Girl using virtual reality headset

Stay updated

Get our latest stories in your inbox every month. Sign up for our newsletters and receive stories of inspiration and discovery, event notifications, breaking news, and more.

Thanks! You'll get a confirmation email shortly.