UK-based Unitary, which uses AI to assess the level of risk and harm in user-generated content uploaded to social media, has raised $8m in additional seed funding in a round led by SongKick’s Ian Hogarth at his VC fund Plural, which he set up with Wise cofounder Taavet Hinrikus.
What does Unitary do?
Unitary was founded in 2019 by Sasha Haco, a mathematician who worked with Steven Hawking on the black hole information paradox during her PhD, and James Thewlis, a computer vision specialist who previously worked with Facebook AI Research.
It uses AI to “read” the context of user-generated videos uploaded to social media platforms and highlight content that could be harmful for particular users or advertisers — without the need for a human moderator. It says its tech can distinguish between a video that glorifies hate speech and documentary footage that aims to raise awareness about hateful content, for example.
As for what defines hate, Unitary fine-tunes its models to identify harmful content based on the specific policies of each of its customers — so, for example, if a user uploaded the same video to Facebook, YouTube and Twitter, Unitary’s AI would measure how harmful that content was according to each of the individual platform’s guidelines. When content is flagged as harmful or risky, the customer is left to take appropriate action.
According to the company, its tech can process up to 25k video frames every second.
So what?
Almost 60% of the global population were using social media in January 2023, uploading years’ worth of footage and content each day. It’s impossible to moderate all of that content manually, and the challenge of keeping platforms free from hate speech is growing. A recent BBC investigation revealed hate is thriving on Twitter under Elon Musk.
That’s opened the door for startups working on the problem and VCs looking to jump on the recent AI bandwagon. According to one report, the content moderation market could be worth $32bn by 2031.
Who invested?
The round was led by Ian Hogarth, CEO of music startup Songkick, at his $250m VC fund Plural, and angels including former Meta executive Carolyn Everson, who will also be joining the board.
What’s next?
The Unitary team say the funding will be used to grow headcount, seal more partnerships and continue working on its open-source tool Detoxify, which can be used to detect written hate speech in the comments section of websites.
The company isn’t the only European startup using AI to moderate content:
-
The UK’s Checkstep, which uses an AI tool and in-house staff to moderate online communities, raised $5m in October last year.
-
France’s Bodyguard.ai, which uses an AI text moderation tool to police online harmful content, cyber bullying and hate speech, raised a €9m Series A last year, according to Dealroom.