Alex Stamos, chief security officer, Facebook:
In our April white paper, “Information Operations and Facebook,” we described the activity that we detected from a sophisticated threat actor that was spreading stolen information about specific political targets in the run-up to the U.S. election and using it to feed press stories that they could then amplify. We took steps to disrupt this activity and reported details to the relevant authorities.
In this white paper, we noted the challenge of attributing threat activity to foreign actors ourselves, but we specifically referenced the assessment of the U.S. government that this actor was tied to Russia’s intelligence services. This was an accurate statement of what we knew about this particular actor at the time, and it appropriately relied on the U.S. intelligence community’s public analysis.
We have been forthcoming at every opportunity about what we know about these information operations. In addition to our white paper, last month we disclosed advertising activity on our platform that we believe is linked to the Internet Research Agency, a different group from the one we described in April. We undertook this research on our own, and we named the group based on our best assessment because we weren’t aware of a comparable public report from the government.
2. Related to the above question: In July 2016, WikiLeaks complained that Facebook was censoring links to a page on its website that hosted the hacked D.N.C. emails. Your chief security officer, Alex Stamos, replied to WikiLeaks (in a tweet that has since been deleted) saying that the issue had “been fixed.” Links to WikiLeaks were subsequently restored. Did Facebook’s security team manually override a tool that flagged these fake accounts as suspicious? If so, who was responsible for the decision to restore access to WikiLeaks, despite having detected a suspicious campaign to promote its stolen documents? Did you notify law enforcement that your security team had intercepted a coordinated influence campaign?
Mr. Stamos: The temporary block of some WikiLeaks links by our automated spam-fighting systems had nothing to do with information operations. It was caused by WikiLeaks posting thousands of raw emails — several of which contained links to malicious phishing and spam sites found in industrywide block lists. We removed the block after we determined that the WikiLeaks links themselves were not harmful.
3. You recently announced you were adding 1,000 human moderators to the team that reviews Facebook ads. How many human ad reviewers did Facebook employ in November 2016? And what percentage of political ads that ran on Facebook during the 2016 election cycle did they review?
Joe Osborne, Facebook spokesman: We don’t usually share the sizes of specific teams at Facebook. Our teams review millions of ads around the world each week, and we use a mix of automated and manual processes. We’re not sharing an exact break-out of the number of manually reviewed political ads.
4. Of the 1,000 human moderators you’re hiring, how many will be based in the United States? Will you be hiring moderators to review ads in non-English languages? What kinds of pre-hire screening will you do to make sure that these moderators are not affiliated with foreign governments, extremist groups, or others looking to influence the American political process?
Mr. Osborne: We are still working through where the moderators will be based, but likely across regions including the U.S., Europe and Asia.
5. You recently told advertisers that new ad campaigns that involved “politics, religion, ethnicity or social issues” would be reviewed by humans before being approved. What guidelines will reviewers be given about which ads to allow and which to reject? Will these guidelines be made public?
6. Your advertising policies allow advertisers to opt out of appearing next to content that involves “debatable social issues.” Which social issues do you define as “debatable,” and how did you make that call? Is your definition of “debatable social issues” globally consistent, or does it vary by region?
7. In countries with regressive social policies, such as criminalizing homosexuality, do you allow the local authorities to determine which issues are considered debatable?
Mr. Osborne: No.
8. Last year, ProPublica found that Facebook advertisers could exclude certain ethnic groups from seeing advertisements about housing, employment and credit, in violation of federal anti-discrimination laws. In response, Facebook announced it would no longer allow ethnic group targeting for those ad categories. Did you consider extending the ban on ethnic group targeting to all ads, including political ads? Did you consider that political campaigns might use ethnic group targeting to suppress voter participation among certain ethnic groups? (Trump campaign officials claim to have used targeted Facebook ads to suppress African-American voters in the weeks leading up to the election.)
9. Did you, at any point leading up to the 2016 election, consider adding disclosures to political ads that made clear who was paying for those ads? If so, why did you decide not to include that feature?
10. You have said you are committed to protecting election integrity and supporting democratic ideals. However, there have been reports that you have built tools to censor speech in certain authoritarian countries, such as China, where you hope to be allowed to operate. How will you choose which elections and democratic processes to protect? When promoting democratic ideals conflicts with your corporate goals, which will you prioritize?
11. Mr. Trump’s digital campaign director, Brad Parscale, has said that Facebook sent “embeds” to work inside the Trump campaign offices and help them use Facebook more efficiently. (You have responded that these offers are standard for political campaigns, and that you “offered identical support to both the Trump and Clinton campaigns.”) What kinds of work did your employees do on behalf of the Trump campaign? Were they involved in writing or editing any of the campaign’s Facebook posts? Were they given page roles or posting rights on any Trump campaign Facebook pages? Were they authorized to report any illegal or suspicious activity they found in the course of their work? If so, did they make any such reports?
12. Your advertiser website lists “success stories” of political campaigns that have used Facebook advertising to increase turnout and win elections. Knowing that Facebook could be used to influence election results, why did you not use a United States presidential election as an occasion to build the proper safeguards to make sure that your system was not gamed by foreign or malicious actors?