Facebook looks to technology to fight fake ads, election meddling, product policy director says
17 November 2017. By Robert Thomason.
Facebook supports making the source and funding of advertisements more transparent, and is planning to improve technologies like artificial intelligence to enforce its own policies and comply with the laws of the countries in which it operates, Facebook’s product policy director said.
The social media giant has been the subject of complaints about suspected Russian actors surreptitiously buying ads and otherwise using Facebook to influence the 2016 US presidential election. Monika Bickert, Facebook's head of product policy and counterterrorism, told a DC audience that the company is increasing its investment to prevent abuse of its system and remove offending content.*
Bickert said that Facebook has 7,000 human reviewers assessing content for compliance with all of Facebook's policies, not just election meddling. But engineers have begun focusing on technical solutions that could flag more suspicious content more quickly.
Much of the work is on counterterrorism, but the firm takes questions about political interference seriously, she said. For example, in the run-up to the French election, Facebook removed 30,000 accounts using improved technical tools.
Artificial intelligence tools and big data analysis performed by the company help to eliminate what Bickert called "weird accounts," although the firm is not as proficient as it hopes to be in identifying bad actors and policy violators. "But I think we will get there," she said.
When Facebook investigated suspicions of interference in the 2016 US elections, she said an existing policy, popularly known as the "Facebook real name policy," was invoked to remove bad accounts. In that probe, Facebook identified 3,000 ads about public policy issues that were purchased between May 2015 and August 2017 by users who purported to be Americans, but upon investigation were deemed to be of another nationality.
"This violated our policy," Bickert said. "You have to be honest about who you are."
Members of the US Congress have introduced legislation called the Honest Ads Act that would require large Web platforms such as Facebook to maintain a public file of people or entities spending more than $500 on political ads, and require the Web sites to make reasonable efforts to prevent foreign interference with US elections.
"We are supportive of more advertising transparency," Bickert said. She said Facebook is researching how to make it clearer who is running an ad. One technique under consideration, she said, is to allow a reader to see the other ads that have been run by someone placing an ad.
Cross jurisdiction legalities
Bickert said that 85 percent of Facebook’s two billion users live outside the US, and the company faces a "lattice work" of data privacy and other online regulation. She described how the social media company faces contradictory laws related to removing or preserving politically or socially controversial speech.
In Germany, a new law called "NetzDG" calls for fines of up to 50 million euros if hate speech or fake news that is "manifestly illegal" is not removed in 24 hours, she said. She said that severe sanctions, which in other countries might even include threats of raiding Facebook offices and jailing employees, lead to a mindset of "if in doubt, take it down." This deprives Facebook users of a great deal of content, she said.
A German user who used a false name sued to maintain the Facebook account in question. The district level judge in Germany ruled for the user, she said.
Polish law protects certain types of speech and requires Web platforms to keep that content online, Bickert said.
*American Bar Association Annual Review of the Field of National Security Law, Washington DC, Nov. 17, 2017.