Misinformation, harassment, grooming and other illegal activity continue to be major issues in the worlds of content moderation and online safety, balancing big problems and illicit activity against equally important, and sometimes conflicting, needs for privacy, data protection and security online. Against that fragmented backdrop, today comes some consolidation and platform growth: ActiveFence, one of the bigger startups building tech for trust and safety teams, has acquired Spectrum Labs, another key startup in the space building AI tools to track online toxicity.
Financial terms of the deal are not being disclosed. Noam Schwartz, the co-founder and CEO of ActiveFence, tells us that the acquisition was predominantly in cash, and that ActiveFence raised money in Q1 of this year that it has never disclosed. (We asked, in multiple ways and from multiple sources, and everyone declined to give details of the amount and the investors in the latest round. But Schwartz confirmed the company’s valuation is now higher than the $500 million of its last round.)
Spectrum Labs was last valued at $137 million per PitchBook data, but its last round of funding was no less than 19 months ago, a $32 million round led by Intel. Other notable backers included Greycroft, Twilio, MunichRe and OurCrowd.
ActiveFence’s acquisition of Spectrum Labs is very timely, and fitting, for the state of the market at the moment.
On one side, you have the swirling currents of the content moderation industry. With We have another major election cycle coming up in the U.S., and while that is once again putting a big focus on the role that social media platforms will play in the misinformation and propaganda games, there is a big question mark over just how all-in companies like Facebook, X (and before it Twitter), YouTube and TikTok will be on trying to police that.
But even without them proactively making choices, some of the decision-making may get taken out of their hands: big online safety bills are making their way through legislatures that could see these companies being required, by law, to do a lot more to make sure their platforms are free of harm.
Yet all that is also still in play: just this week in the UK, where the country is working through its Online Safety Bill, the government found itself back-pedalling on an earlier requirement for messaging apps to implement encryption-breaking surveillance tools: whether it was even technologically possible, it raised many, many questions about security and data protection. Similar tug-of-wars are playing out elsewhere.
On the other side, the acquisition is a classic development in what is a maturing market, but also one operating with an economically constricted environment where investors (and customers) are making difficult choices when trying to back “winners.”
We’ve written about ActiveFence previously: in 2021 the company, based in New York but with roots out of Israel, came out of stealth with $100 million in funding that included another undisclosed round that it had kept quiet for years. (Its investors include Highland Europe, CRV, Telefonica, Norwest Venture Partners and more).
The startup’s core proposition has long been to develop tools, and a platform, for trust and safety teams to help them do their complicated jobs more easily. The evolution of how the market will look longer term is likely to be, Schwartz said, “somewhat similar to the likely evolution of cybersecurity” — which is to say a wide-ranging toolkit of services covering the many different aspects of how to monitor and shut down harmful and illegal content, a one-stop shop rather than multiple point solutions.
It says its customers are some of the biggest companies in social media, gaming and e-commerce, in all covering over 3 billion consumers across more than 100 languages.
Spectrum Labs, based out of Miami, is effectively coming on to become one more part of that mix.
Its own business had a lot of momentum before the deal. Co-founders Justin Davis and Josh Newman started the company having worked for years in marketing tech, and moved by what they saw as a truly dire state of affairs in the world of online toxicity, they saw an opportunity to take their expertise in big data analytics and tagging and tracking content for brands and corporates and apply it to fixing that problem. But the acquisition will not only bring technology and talent to ActiveFence, but also customers: some of the companies using Spectrum’s products included Riot Games, Grindr, and The Meet Group.
“We’re building a Trust & Safety platform to support the needs of one of the most complicated and underserved segments in technology. Acquiring point solutions like SpectrumLabs is a key piece of the puzzle,” said Schwartz. “The decision to acquire the company was based on multiple strategic considerations, and to put it succinctly, it’s both the customers and the tech that drove our decision. Spectrum has cultivated a loyal customer base… that complements our existing clientele and presents us with an opportunity to provide enhanced offerings and value to a broader market.
“On the technology side, Spectrum has developed certain proprietary technologies and processes that supplement ours from several angles. Specifically, some features of their tech and unique capabilities around text-align closely with our vision for the future, and by integrating them into our existing comprehensive solutions, we can accelerate our product roadmap and deliver better solutions to our combined customer base. This acquisition isn’t just about short-term gains; it’s about setting the stage for long-term innovation and growth, in a Trust & Safety arena that we believe is destined to be dominated by platforms, like us, rather than a large number of point solutions and smaller vendors.”
Yet funding being what it is these days, you can definitely see a pattern emerging around runway, better economies of scale, and more M&A on the table.
ActiveFence already leans on artificial intelligence in its approach, and longer term that will continue to be a greater part of the mix.
“GenAI has, in one motion, made harmful content detection more difficult and reduced the challenge for bad actors to create sophisticated disinformation campaigns,” Schwartz said, citing Meta’s latest harmful content report to illustrate that, “and more elusive abusive materials using AI models that perfectly imitate real people. There is a dangerous trend of democratization of abusive online behavior and ActiveFence is at the forefront of the battle against such bad actors.”
But on the defensive side, he added, it’s also getting more prominent, a large part of how organisations can scan and process huge swathes of content to sense out patterns as well as anomalies, and identify actors. “Large Language Models are one piece of the puzzle and the ability to leverage these models to detect content is a wonderful way to advance our abilities every day. Additionally, by training genAI models with platform- and policy-specific content, we can create more customized content policies – a complete game changer in content moderation.”
Longer term, the message seems to be that there will be more M&A coming as a quicker route to expanding the product to better fight the harmful content creep — and there are many interesting startups still independent that might be targets — but ActiveFence itself, he said, is not for sale.
“Given market conditions, I expect more M&A activities to take place in the future,” he said, “[but] as for ActiveFence, while we receive interest on an ongoing basis, our primary focus remains on enhancing our capabilities and serving our customers.”