Rupert Small calls superhuman-level AI “the ticking time-bomb on the centre of the web”.
The researcher-turned-founder — who beforehand labored at metaverse unicorn Inconceivable — began his firm Egregious AI in 2023 to guard people from the “spectacular bullshit” pumped out day by day by autonomous AI bots and brokers.
As AI will get higher at imitating people on-line, the danger of unhealthy actors deploying the tech at scale for market manipulation, fraud and big misinformation campaigns grows. The risk is giving rise to startups using AI to battle AI.
Europe trails the US on this budding area of interest. Throughout the pond, there have been huge rounds raised by firms preventing malicious AI, reminiscent of Alethea (raised $30m) and Blackbird AI ($30.6m). San Francisco-based Human.org, which earlier this month raised $7.3m, claims to be the primary firm trying to confirm whether or not an autonomous AI agent represents an actual particular person.
Funding in Europe is smaller, however listed here are seven startups which have raised VC capital to guard web customers from AI scams, copyright theft, disinformation and deepfakes (and one startup that could possibly be on the fundraising path quickly).
IdentifAI, Milan
This Milan startup, which raised a €2.2m seed spherical led by Italian VC United Ventures in 2024, is creating what it calls “(de)generative AI” to establish bots. Founder Marco Ramilli says it’s more and more tough to differentiate between pc and human creations: “I imagine it’s our proper to learn, always, about what we use to nourish our minds, which in flip shapes our opinions.”
Ceartas, Dublin
This startup makes use of bots to seek out copyrighted content material then routinely requests Google take down the hyperlinks. Many content material creators use the service to take away deepfake pornography from the web (Ceartas is an official “security associate” of subscriber platform OnlyFans). Ceartas, which launched in 2021, raised €4m final 12 months from lead investor Earlybird Enterprise Capital, with participation from Upside VC, established by Youtube creators The Sidemen and a raft of angel buyers.
Egregious AI, London
This startup has punchy ideas on the dangers posed by AI. In response to Egregious, “Superhuman AI is essentially the most disruptive invention in human historical past.” In January, the startup raised a $1m pre-seed spherical, led by Gas Ventures and Oxford Capital, to develop a defensive AI arsenal for the web.
Refute, London
This startup, run by Tom Garnett and Vlad Galu, detects and responds to disinformation on behalf of firms. It raised a £2.3m pre-seed spherical led by buyers Playfair and Episode 1 in December 2024.
Osavul, Kyiv
A Ukrainian startup shaped after Russian missiles destroyed the cofounder Dmytro Bilash’s flat in Kyiv. Headed by Bilash and Dmytro Pleshakov, Osavul makes use of AI to sift out “false narratives” on the Russian-language web. The corporate’s tech also can detect if data is being pushed in a coordinated method on-line. The corporate raised €2.8m final 12 months from German VC 42CAP, US government-backed u.ventures and Ukrainian fund SMRK.
DuckDuckGoose, Delft
This Dutch startup — named for the kids’s sport during which one identifies a goose amongst geese — raised €1.3m in 2024 to confirm whether or not video, photos or audio have been manipulated. The corporate’s “explainable AI” highlights particular areas of media which are distorted.
DarkForest, London
This effort by AI researcher Mustafa Yasir, which hasn’t sought any VC cash but, needs to offer web goers higher management over how AI bots entry their content material. With a number of traces of code, DarkForest will redirect bots to a web page of the person’s alternative or a paywall demanding AI firms implement particular protocols. Yasir, on sabbatical from The Alan Turing Institute in London, is creating his concept on the new hacker lodge in Espoo.
Logically, Brighouse
Yorkshire-based startup Logically has created a collection of AI-based merchandise designed to assist folks, companies and governments establish false tales and narratives on the web, stopping their unfold and monitoring sources of misinformation. Earlier this month, Sifted completely revealed the corporate had laid off dozens of workers as a part of a company-wide effort to chop prices.
