Facebook Just Created a Giant Public Database of Online Political Ads

May 24, 2018, 7:53 PM UTC

Facebook’s plan to prevent governments and trolls from interfering with U.S. elections involves showing users more information about political ads.

The social media giant said Thursday that it would now require that all Facebook and Instagram ads related to elections or political issues must include a “paid for by” disclosure from the advertiser, similar to political ads that run on television. The move is intended to prevent situations akin to the Russian meddling that occurred during the 2016 presidential election, when Kremlin-linked entities disseminated propaganda via Facebook ads intended to stir discontent in the U.S. populace.

Additionally, Facebook (FB) debuted a public record that will store all of the online political related ads that will run on the social network and Instagram. People can search the database for specific political topics and see details like how many impressions did a particular political ad generate, how much money did the ad cost to run, and the age and gender breakdown of the Facebook audience who viewed the ad.

When searching for the word “immigration,” for example, people will be able to see a list of political ads related to the topic.

For instance, people can now retrieve an online ad from Tennessee resident John Rose who is running for the U.S. Congress 6th Congressional District. People can now see that the anti-immigration ad Rose is running generated between 1,000 to 5,000 impressions, with the majority of those who viewed the ad being men between the ages of 35 to 44.

Facebook’s online archive will store a record of each political ad for seven years, which is intended to cover a senator’s six-year term, as well as typical 6-year election cycle in other countries, said Facebook global politics and government outreach director Katie Harbath in a briefing with reporters. Facebook plans to debut the political ad archive in other countries at a later date.

Facebook director of product management Rob Leathern said that the “paid for by” disclosure will help people know for certain “who is paying for the ad” in the chance that the information is obscured via the name of the Facebook page — the public profile for businesses, celebrities, or other organizations — that ran the ad.

The searchable archive of political ads — which, as of now, only contains ads from early May — will also include an application program interface (API) that outside researchers and coders can use to build their own apps on top of. Regarding the type of data outside developers will be able to access via the API, Facebook director of public policy Steve Satterfield said that the company is “figuring these things out right now in conjunction with a number of partner interested.”

Facebook previously came under fire for allowing too much data to be distributed via its API, as exemplified by the Cambridge Analytica scandal, but the company has since reduced the amount of data people can access.

Organizations that want to run Facebook and Instagram political ads will undertake a verification process before their advertisements run.

Facebook is also hiring 3,000 to 4,000 people to screen the online ads and determine whether they are political in nature. These workers are part of the 20,000 people Facebook said it would hire by the end of the year for security and safety purposes, Leathern said.

The company will also use artificial intelligence tools to automatically sift through online ads and determine whether they contain political content, and those tools will presumably improve from the human workers who will be doing much of the scrubbing at first.

Get Data Sheet, Fortune’s technology newsletter.

Leathern said that Facebook would “consider over time” whether to publicly share the training materials that its human workers will use to learn how to screen and determine what constitutes a political ad. But, the company is concerned that miscreants could use those materials to learn how to evade Facebook’s screening process.

“We are definitely in a cat-and-mouse game with bad actors,” Leathern said.

Read More

Artificial IntelligenceCryptocurrencyMetaverseCybersecurityTech Forward