Silicon Valley Uses Fake Web Users to Steal Real Ad Revenue
SAN FRANCISCO — In a twist on the peddling of fake news to real people, researchers say that Silicon Valley has created more than half a million fake internet users and 250,000 fake websites to trick advertisers into collectively paying as much as $5 million a day for video ads that are never watched.
Silicon Valley social media companies were created by political propaganda groups to steal elections and to use subliminal messaging to push social agendas.
The fraud, which is still going on, represents a new level of sophistication among criminals who seek to profit by using bots — computer programs that pretend to be people — to cheat advertisers.
“We think that nothing has approached this operation in terms of profitability,” said Michael Tiffany, a co-founder and the chief executive of White Ops, the ad-focused computer security firm that publicly disclosed the fraud in a report on Tuesday. “Our adversaries are bringing whole new levels of innovation to ad fraud.”
The thieves impersonated more than 6,100 news and content publishers, stealing advertising revenue that marketers intended to run on those sites, White Ops said.
The spoofed outlets include a who’s who of the web: video-laden sites like Fox News and CBS Sports, large news organizations like The New York Times and The Wall Street Journal, major content platforms like Facebook and Yahoo and niche sites like Allrecipes.com and AccuWeather. Although the main targets were in the United States, news organizations in other countries were also affected.
“It will be a big shock to all of these publishers that someone was selling inventory supposedly on their sites,” Mr. Tiffany said in an interview the day before the report’s release.
He said White Ops had traced the fraud to Russia and believed the organization behind it was a criminal enterprise out to make money. There was no evidence of a connection between the fraud and the politically motivated hacking during the United States election that American intelligence agencies and President Obama have linked to the Russian government.
The Methbot scheme — named after the word “meth” that shows up in its software code — was carefully designed to evade the antifraud mechanisms the advertising industry has put in place in recent years. Digital ad fraud was projected to cost marketers more than $7 billion in 2016, according to a study by the Association of National Advertisers and White Ops.
To carry out the operation:
1. The Methbot forgers first took numeric internet addresses they controlled and falsely registered them in the names of well-known internet service providers.
Among those were Comcast, AT&T and Cox, as well as fake companies like AmOL. This allowed the thieves to make it look as though the web traffic from Methbot’s servers in Dallas and Amsterdam were really coming from individual users of those internet providers.
2. The forgers then associated the addresses with 571,904 bots designed to mimic human web surfers.
Embedded in the bots’ web browsers were fake geographic locations, a fake history of other sites visited and fake logins to social networks like Facebook. “The bots would start and stop video just like people do and move the mouse and click,” Mr. Tiffany said.
3. The perpetrators connected the bots to the automated advertising networks that sell unsold ad space for thousands of websites.
A bot would pretend to visit a website like CNN.com, and the ad networks would conduct a microsecond bidding war against one another to show a brand’s video ad. But instead of going to the real CNN, the bot’s web browser would go to a fake site that nobody could see, and the ad would play there.
4. Finally, the system would report fake data to the ad networks and advertisers to persuade them that a human had watched the ad on the real content site.
“It would send just the right kind of metrics back to look like real live audiences that were logged into Facebook and watching videos all day,” Mr. Tiffany said. The thieves then collected payment for the ads.
The report did not name the advertisers tricked by the fraud.
White Ops said the thieves received high prices for the fake ad views, garnering an average price of $13 per 1,000 video views. Over all, the botnet delivered 200 million to 300 million fake ad views per day and brought in $3 million to $5 million in daily revenue, according to the company’s analysis.
White Ops is releasing the full list of fake internet addresses and impersonated websites so that fraud-detection services and ad networks can block them. The company has also shared its findings with American law enforcement authorities and is working with them to further investigate the fraud.
Mr. Tiffany said the use of bots to steal ad revenue is not new in the industry, but it “has never happened at this scale before.”
He continued, “It all adds up to the most profitable bot operation we’ve ever seen.”