Spectrum Labs raises $32M for AI-based content moderation that monitors billions of conversations daily for toxicity


Even though the epidemic has been ongoing for two years, online discussions are still the primary form of engagement for many of us daily. We collectively engage in billions of them. However, as many of us have experienced, not every one of those encounters is a gleaming gem of happiness. Spectrum Labs, a business that delivers artificial intelligence technologies to platform providers to identify and stop down harmful transactions in real-time (particularly within 20 milliseconds or less), said today that it had raised $32 million in investment. With the funds, it plans on investing in its technology to further expand its consumer business while also forging ahead in a new area, providing services to enterprises for their internal and customer-facing conversations, not only providing a way to help detect when toxicity is creeping into exchanges, but also providing an audit trail for the activity to support more comprehensive trust and safety tracking and initiative tracking.

“We hope to be the leaders in a language where civility is important,” CEO Justin Davis stated in an interview with Business Insider.

A total of $1 billion has been raised in a round headed by Intel Capital with participation from Munich Re Ventures, Gaingels Ventures, OurCrowd, and Harris Barton, as well as prior backers such as Wing Venture Capital, Greycroft, Ridge Ventures, Superset, and Global Founders Capital. Greycroft led Spectrum’s last round of funding, which received $10 million in September 2020, bringing the company’s total budget to $46 million.

Davis, who co-founded the firm with Josh Newman (its chief technology officer), noted that although Spectrum Labs does not disclose its worth, its current business size testifies to how well it has done.

Spectrum Labs collaborates with just over 20 major platforms, including social networking companies Pinterest and The Meet Group, dating site Grindr, Jimmy Wales’ entertainment wiki Fandom, video game developer Riot Games, and e-learning platform Udemy. These platforms, in turn, have millions of customers who send billions of messages to one another every day, either in open chat rooms or in more direct, private conversations.

Incorporating natural language into its technology allows it to function in real-time on both text-based interactions as well as audio-based interactions.

In addition, Davis points out that Spectrum’s audio work is “read” as audio rather than being transcribed to text first, giving its customers a significant head start on responding to the activity and counteracting what Davis refers to as “The Wild West nature of voice,” which refers to how slow responses can be when using other technologies, such as those used by traditional media platforms: a platform must wait for users to flag iffy content, then the platform must find that audio in the transcriptions, and then it can take

This is especially essential since voice-based services are becoming more popular, thanks to the emergence of podcasting and services like Clubhouse and Spaces on Twitter, among other things.

Spectrum analyzes these interactions for hazardous material, whether text or voice and does so using more than 40 behavior profiles that it developed in cooperation with experts and academics across the globe. It continues to refine as it ingests more data from across the web. They include aspects such as harassment, hate speech, violent extremism, frauds, grooming, unlawful soliciting, doing, and other related topics. Davis informs me that it presently supports scanning in over 40 languages and could function with any language in the future. Davis also tells me that there is no language restriction at this time.

We can potentially cover every language in a couple of weeks, according to Mr. Chen.

On-line toxic behavior has manifested itself most visibly among consumers in the form of open-forum and private online bullying and hate speech, as well as other illegal activity. In this area, Spectrum Labs will continue to research and develop technology to detect ever more complex and sophisticated approaches by bad actors. In addition to developing controls and tools for a platform’s trust and safety team, Spectrum Labs will look at methods to enhance how consumers themselves may have a part in determining what they do and do not want to view. Since the other side of the coin is that they can also be accused of censorship, which is still a hotly debated topic today, this is a complex area to navigate. One of the reasons that toxicity has gotten out of hand, according to some, is that platforms have traditionally preferred to take a hands-off, free speech approach and not meddle in content.

According to Davis, there is a “naturally occurring friction” between what the policy entails and what consumers want and are ready to tolerate. His organization believes that the role of a platform is to “keep the worst of the worst off the platform, but also to empower consumers with controls to make selections about what they want to see over time.”

In addition to that, Spectrum intends to expand its enterprise services offerings.

It’s an intriguing opportunity in the enterprise because it includes not only how people within a company communicate with one another (which essentially could take the form of the consumer-facing services that Spectrum Labs already provides) but also how a company interacts with the outside world in areas such as sales, customer service, and marketing, and then leveraging the information that Spectrum Labs gathers in its analytics to potentially alter how each of those areas interacts with the outside world.

Indeed, this is not a market area that has gone unnoticed in the past. Spectrum’s rivals in this field will include Aware, a company in the conversation monitoring space that focuses only on the business market. The company L1ght, on the other hand, is a rival in the consumer market.

There will, without a doubt, be others after that. Previously, we said that the founders and founding staff were recruited from Krux, a marketing technology startup bought by Salesforce. This is still true (where they worked before leaving to found Spectrum Labs). Not only because Salesforce is building out a comprehensive toolset to help companies run their businesses more efficiently in general, not just CRM, but also because Bret Taylor, who founded another social network and used to be the CTO of Facebook, is now helping to run the show and may well have an exceptionally informed grasp on how communications forums can be used and abused, I wouldn’t be surprised to see Salesforce taking a more exciting role in this area in the future.

For the time being, Intel will participate in this round as a strategic investor to address both consumer and enterprise challenges, according to Davis. The plan is to integrate Spectrum Labs’ technology into Intel’s chip designs, which will increase the speed at which it operates even further. Intel will be able to use this as a unique selling point with prospective hardware customers, who are placing a higher priority on trust and safety issues themselves due to this integration.

“We are confident in Spectrum Labs’ Natural Language technology.” According to Mark Rostick, vice president and senior managing director at Intel Capital, “understanding technology has the potential to become the underlying platform that supports the trust efforts of thousands of firms across the globe.” We believe that as digital trust and ethical operations become more prevalent for enterprises to distinguish themselves, there is an enormous potential to integrate a Trust & Safety technology layer into corporate operations.


Please enter your comment!
Please enter your name here