Social

Fake news, and the identification and eradication of it, has long been thought of as the purview of social media platforms, where a lot of that tends to be shared. Today, one of the more ambitious tech startups in the field of fighting fake news is getting acquired — not by a social media platform, but by a player in of the other parties that stands to gain and lose a lot from misinformation, or at least bad press: PR. Factmata — founded by AI specialists with the aim of building an engine to detect when fake news and other false information is shared online, but which had more recently turned to using its tech for social analytics — is being acquired by Cision, a provider of media monitoring and distribution services and products for the public relations industry with some 100,000 customers. The financial terms of the deal are not being disclosed, both companies tell me. Cision — now privately held after being taken off the public market at a $2.74 billion valuation in 2019 — is one of the bigger companies in the field of public relations services, owning brands like PR Newswire and managing databases provided to PR professionals to better target their pitches (or not, as the case often seems to be). Over time it has expanded beyond engagement and deeper into areas like media monitoring, and it has made a number of acquisitions to bolster that, such as its 2021 acquisition of Brandwatch in the UK for $450 million. It’s very unlikely that this deal was anything close to that. Factmata had raised around $4 million in total, with its investors including a number of high-profile individuals in the world of news, online information and media, including Biz Stone, Craig Newmark and Mark Cuban. The company has a staff of just seven people, and it sounds like will include Factmata’s IP as well as all of of them, including CEO Antony Cousins, who are all joining Cision to build out its existing tools as part of a bigger automation and AI play that Cision is pursuing. The acquisition is notable for a couple of reasons. The first of these is that it gives us a moment to take a pulse check on social media moderation, and how that is being tackled. Fact checking and social media moderation have been hot topics for years. But with layoffs at big platforms like Twitter and Facebook hitting engineers and content moderation teams, a big question has emerged: how effective will these and other companies be at parsing and responding to the waves of content spurred by elections and other big events — especially since their ability to stem the tide of violent, harassing, and misleading posts was never perfect to begin with? For better or worse, that has created an opening — or a responsibility, depending on how you look at it: organizations or individuals who want to understand how the world is talking about them, and to stem the tide of bad conversations, are going out there and finding out for themselves. For its part, Factmata had high hopes when it was first founded by Dhruv Gulati, Sebastian Riedel and Andreas Vlachos (none of whom are working with the company now: Gulati who had been the CEO and then chairman is at Onfido; Riedel — who also founded and sold another news detection startup Bloomsbury AI to Facebook — is at Deepmind, and Vlachos is an academic at Cambridge). It set out to build an engine to find and respond to fake news automatically, with the potential users including not just those social media platforms but also consumers. At one point the company cut a deal with the creators of AdBlock Plus to take investment from them and to take on the operation of their Chrome plug-in Trusted News: the idea was that this would help Factmata develop its go-to-market strategy, by helping it ingest more fake news (and trusted news) data, but also to have a direct line to users. But Cousins tells me that this vision evolved over time. That was in part because it found that targeting a few social media companies was not as scalable as a business model as targeting the wider world of brands and businesses, Cousins said. It was also because Factmata found that there was too much nuance in a lot of content, and ultimately, while an AI system is much better at surfacing clusters of activity and evolving trends, or narratives as the company describes them, a human in the loop, he said, is needed to determine which of these are actionable and which are false flags. “We focused on building tech that could find a narrative but then let users judge for themselves if that narrative is worth watching,” he said, describing it as a “picks and shovels” approach. “It’s scalable and appropriate to have a human making the ultimate decision.” That is what Factmata discovered, and that’s what a lot of others in the field also believe is likely the right route forward. Now, it will be interesting to see how that plays out for companies that are removing content moderation teams now due to cost cutting. Another reason why Factmata’s acquisition is notable is because of the wider startup context. From what we understand, the startup was finding it a challenge to raise money in the current climate, given that it was seeing some growth but not enough, and it was approaching the end of its cash reserves. The company has been working with a number of high-profile companies and their agencies to incorporate its automatic “narrative” detection into their wider monitoring activities, and by April of this year it was “halfway to break even” according to Cousins. “But that was just not enough evidence for VCs,” he continued. “They needed to see evidence of six to nine months of growth. So it wasn’t enough, and we didn’t have the runway for another year.”