Giskard raises $1.5 million to secure artificial intelligence

“An AI error can have catastrophic ethical consequences and cause a company to lose a lot of money. The Zillow company thus lost 569 million euros because of a faulty algorithm, which estimated the value of real estate.recalls Alex Combessie, co-founder of the start-up Giskard, alongside Jean-Marie John-Mathews and Andrei Avtomonov. To avoid errors, these three experts – who previously developed critical AI systems at Dataiku and Thales – have developed a quality assurance platform for artificial intelligence algorithms. A tool that also makes it possible to adapt to European regulations on AI, which will soon apply fines of 6% of turnover in the event of incidents on high-risk algorithms. Because the problem is not uncommon. In total, more than 2,000 incidents have been reported to the AI ​​Incident Database since 2019.

Investing in new technical areas

For Giskard’s tool to be able to limit the risks for all kinds of AI, the company created a year ago announces its first fundraising. A $1.5 million financing round led by Elaia with the participation of Bessemer Venture Partners and recognized business angels: “With these funds, we want to secure the AIs that work on the image, which we are not doing yet, and invest in new technical areas to expand into fraud detection, for example”, explains Alex Combessie. In parallel, Giskard also plans to focus on vision, time series and generative AI algorithms. To do this, the company of 8 employees wants to almost double its workforce by recruiting 7 people. Recruitments which aim to expand its tech department but which will also make it possible to accelerate the marketing of the platform. Giskard targets artificial intelligence developers and already works with companies such as Webedia, CGI, Citibeats, Altaroad, or Unifai.

“It is very frustrating for engineers because it is complex to guarantee the security of an algorithm. And that can have serious consequences., completes the leader. In the case of autonomous vehicles, this can indeed cause accidents or even the loss of human life. “The more ubiquitous AI becomes, the more impact it will have on our lives. This increases the need to test algorithms”says Sébastien Lefebvre, Partner at Elaia.

Prevent the risk of incidents in production

The platform now offers two features. First, it visualizes and explains AI algorithms to identify errors. Then, the tool offers the possibility to build automated tests on the algorithms to measure critical quality criteria, such as ethics and robustness. Companies can thus prevent the risk of incidents in production and implement a process of continuous evaluation.

We want to say thanks to the writer of this short article for this amazing web content

Giskard raises $1.5 million to secure artificial intelligence

Discover our social media profiles , as well as other pages that are related to them.