China's hidden targeting of Donald Trump one reason Facebook needs new cybersecurity approach
Six months out from election 2020, Facebook is awash with coronavirus-themed paid advertising — but the American public has been left in the dark about who is paying for these ads and how they are being targeted. Facebook has taken steps toward transparency by voluntarily disclosing political ad content and data through its online Ad Library. But the library is complicated to use, untold numbers of ads are missing and a significant element is lacking: adequate information on how ads are directed toward specific demographics and groups of people.
For the public to be better informed about who is trying to influence their votes and why, we need to embrace new approaches for transparency. Tried and true techniques from the practice of cybersecurity can help us strengthen democracy by revealing vulnerabilities in disclosure of online political ads.
Transparency is needed
In the 2010 Citizens United decision, U.S. Supreme Court Justice Anthony Kennedy famously affirmed the salutary effect of sunlight on democracy: “With the advent of the internet, prompt disclosure of expenditures can provide shareholders and citizens with the information needed to hold corporations and elected officials accountable for their positions. ... This transparency enables the electorate to make informed decisions and give proper weight to different speakers and messages.”
But paid political content on Facebook, Google, Snapchat and other social media platforms comes in many sizes and disguises. Laws and regulations on disclosure of political spending have not adapted to our online world, with its walled garden communities that each operate under their own opaque algorithms designed to meet different business models.
In one recent stark example, Vice reported that the Chinese government, via state-run media organizations, has run a series of ads that blame President Donald Trump for the spread of coronavirus, but that until now these had not been tagged as “political” by Facebook.
New York University’s research showed that one of these state agencies, Xinhua News Agency, regularly posts paid political content not disclosed as such by Facebook.
Facebook’s own rules prohibit foreign spending on ads of social, national or electoral importance. But these ads may slip through because they are ostensibly produced by media organizations, which have long qualified for exemption from political spending disclosure rules — although in this case these are state-owned, foreign entities. Chinese media agencies don’t advertise on mass media such as U.S. television stations or highway billboards, but on social media, it’s easy to slip in.
Social media is not a source of peril: It's a powerful tool
Much domestic “dark money” political spending also eludes Facebook’s disclosure. In March, a nonprofit group called Fellow Americans posted an ad juxtaposing President Trump’s statements playing down the threat of COVID-19 with visuals showing exponential growth of cases. Fellow Americans has reported $181,000 in spending on anti-Trump media buys of any type, according to OpenSecrets' analysis of recent Federal Election Commission filings. However, disclosure by Google and Facebook of Fellow American digital ads total just $36,193, or roughly 20% of the total. Meanwhile recent research on TV ad buys in the presidential race done in partnership between OpenSecrets and the Wesleyan Media Project reveal no spending by Fellow Americans on TV ads.
The discrepancy shows shortcomings of disclosure by social media platforms. And while totals are reported to the FEC, these include just a dollar amount, not actual copies of ads or information about how they were targeted at specific audiences. Maybe they were underreported by these platforms, or perhaps they ran on online platforms that don’t disclose political ads, such as Hulu or Pandora.
It's time for a new approach
What can we do in the face of regulatory failure for online political ad disclosure, as well as shortcomings in voluntary disclosure? We need to augment methods traditionally used by journalists and campaign finance researchers with new approaches adopted from the field of cybersecurity. In a typical cybersecurity analysis, computer science experts deliberately attempt to find vulnerabilities in a particular online system, whether it’s the public site of a large bank or government institution. Once they find these vulnerabilities — maybe it’s a hole that allows spam messages to get out via a group email list or a mistake that exposes personal information of customers — organizations have information they can use to change their policies and better serve their public.
One of us is a cybersecurity researcher at New York University, and has used such methods to show how Facebook failed to disclose crucial information about tens of thousands of political ads in 2018 in violation of the platform’s own policies. These ran the gamut from lack of standardization of “disclosure strings” supposed to contain names of sponsors of political ads to examples of “inauthentic communities,” or clusters of similar ads, designed to look like they are coming from different, unrelated groups.
The other of us is director of the Center for Responsive Politics, which has been collecting, coding and analyzing data on political spending for several decades via official filings with the FEC, Federal Communications Commission, IRS and other government agencies. Cybersecurity methods reveal potential problems with online political disclosure. Making connections between this information and political influence databases of lobbyists and campaign donors will deepen our understanding.
Facebook: We shouldn't become the gatekeeper of truth on candidate ads
In the past, social media platforms have thwarted researchers and journalists who probe at their vulnerabilities. Last year, Facebook shut down ProPublica’s political ad tracker, which allowed users to volunteer their anonymized data on which Facebook ads they were seeing. But our work has shown conclusively that voluntary disclosure is severely lacking and does not paint a comprehensive picture of political influence online.
Meanwhile, the newly announced Facebook Oversight Board will primarily deal with appeals to reinstate content that has been removed from the site, not how well Facebook is disclosing political ads.
As we head toward election 2020 in the midst of a pandemic, many forces will try to game online platforms to the detriment of our open political system. Social media sites could use help from us and other researchers working to bring sunlight to political ads.
Laura Edelson is a Ph.D. candidate at New York University’s Tandon School of Engineering. Sheila Krumholz is executive director of the Center for Responsive Politics (OpenSecrets.org). Follow her on Twitter: @skrmhlz
You can read diverse opinions from our Board of Contributors and other writers on the Opinion front page, on Twitter @usatodayopinion and in our daily Opinion newsletter. To respond to a column, submit a comment to letters@usatoday.com.
This article originally appeared on USA TODAY: Democracy in danger: Facebook needs a new approach to cybersecurity
