Founded by ex-NATO AI and cybersecurity experts, Sentinel is backed by seasoned angel investors, such as Jaan Tallinn (Skype), Taavet Hinrikus (TransferWise), Ragnar Sass and Martin Henk (Pipedrive). The investor lineup includes United Angels VC, an early stage VC fund in the Baltics and Nordics with notable investments in tech startups like Bolt, Monese, Starship and Veriff.
“Sentinel is an example of developing defensive technology that can mitigate a potentially destabilising race,” commented on his latest investment Jaan Tallinn, the founding engineer of Skype, now dedicated to AI alignment and the mitigation of existential risks from advanced technology.
Sentinel stands for true and trustful information
Several studies show, that realistic-looking images and videos produced with AI that portray someone doing or saying something that never actually happened are growingly being used in politics, information operations and elsewhere.
According to an Oxford University research paper, recent years have witnessed a massive growth in the number of countries launching disinformation campaigns, with 70 countries on the blacklist in 2019, compared to 28 just two years earlier. Inexpensive distribution channels such as social media rapidly increase the capability to manipulate citizens, damage the economy and disrupt democratic processes.
Sentinel co-founder and CEO Johannes Tammekänd, who has previously worked at NATO Cooperative Cyber Defence Centre of Excellence, finds that advanced deepfakes are now beyond identifiable by the naked eye and ear, sowing doubt on authentic media content.
„When any video can be a deepfake, instead of being deceived, people may grow to distrust a large part of video and audio recordings. Furthermore, neglecting the authenticity of the content altogether, people may become susceptible to believing only what confirms their biases,” Tammekänd noted. “In the long term, the majority of digital information will be automated and produced by AI, which necessitates a trust layer for the Internet to protect people against information warfare,“ Tammekänd added.
Today, sophisticated adversaries can produce deepfakes at scale, using only a single selfie. A recent study by Sentinel shows an exponential growth in deepfakes found in the wild (anywhere on the Internet), totalling to more than 145,000 in 2020, which indicates a ninefold year-on-year growth. The authors of the study bring out the heightened probability that some of the upcoming key political events, such as 2020 U.S. presidential elections, will see the deployment of AI-enabled information operations through deepfake video, text or audio in an attempt to influence the outcome.
Sentinel works with some of the leading International media outlets, governments and defence agencies, including European External Action Service, the combined foreign and defence ministry of the EU. The company’s early clients include the Estonian Government – one of the most advanced in e-governance, digital identity and cybersecurity. Estonia is pledging to become a flagship country in countering information warfare and has involved Sentinel in its latest innovation project called Accelerate Estonia.