Yahoo Finance LinkedIn says it took down millions of fake accounts as experts warn it has become a 'gold mine' for spies
LinkedIn said that it removed more than 21m fake accounts in the first half of 2019, amid mounting evidence that the business-focused social network has become a battleground for duelling spies.
The Microsoft-owned company's first moderation report, published on Thursday, revealed that 19.5m fake accounts had been automatically detected at registration by its AI systems, with a further 2m "restricted" after being created.
It also showed the biggest ever rise in government requests for LinkedIn users' data, which increased by 46pc from the previous reporting period to a record high of 362.
Many of those of those appeared to be "national security" orders from the US government, which US tech firms are allowed to disclose without giving exact numbers. LinkedIn said that of the 1,170 total accounts that US officials sought to tap, between 250 and 499 were swept up in such orders.
Rob Hallman, LinkedIn's vice president of legal affairs for products and privacy, said that the company "carefully reviews each government request before taking any action", and notifies their targets whenever the law allows it.
It comes after multiple US government officials and security experts warned that LinkedIn is being used by foreign spies to gather intelligence and worm their way into the good graces of well-connected sources.
William Evanina, director of the US National Counterintelligence and Security Centre, warned in August that China is using LinkedIn "on a mass scale", while Clint Watts, a former FBI agent, told the US TV network CNBC this month that it was a "goldmine" for spies.
In May, a retired CIA agent named Kevin Mallory was jailed for 20 years for selling state secrets to the Chinese government, a scheme which first began with a LinkedIn message from a Chinese agent posting as a recruiter.
LinkedIn's hard-nosed, self-promoting culture, in which users often fire off salvos of unsolicited messages to potential contacts and treat enlarging their professional network as a prime directive, provides a welcoming environment for spooks hoping to cultivate new assets.
An investigation by the Associated Press found that prominent residents of Washington DC had unwittingly "connected" with an entirely fake profile whose face appeared to be generated by a computer program. Paul Winfree, a former White House official, admitted: "I literally accept every friend request that I get."
LinkedIn also revealed that its biggest category of rule-breaking content was harassment, with 16,626 instances removed. The next largest was "adult" content, of which 11,019 instances were removed, with just a few thousand instances of hate speech and graphic material.
The number of fake accounts paled in comparison to that reported by Facebook, which said that it removed a staggering 3.7bn fake accounts during the same period. That number was higher than the actual accounts reported in its most recent financial results.
