Adversarial AI – Who Monitors Artificial Intelligence?
This post is also available in: 
עברית (Hebrew)
Today, more and more systems are enveloping us in our daily lives. From autonomous cars to charging your phone, all these actions create information and access for AI to utilize. Everywhere we go we are being identified. At the airport, on the road, at supermarkets and more via identification cameras and other technologies.
As opposed to script based systems, identification systems are built on artificial intelligence. These AI systems gather millions and millions of photographs and create a template for recognition. Each combination of pixels will be worth a value, given a meaning. However, the accuracy of these systems is never 100% and mistakes are bound to happen. In fact, many already have.
Now imagine if that same information was poisoned and the damage it can cause. If a hacker would get access to such a database and be able to change eve n a few small pixels, the system would be poised and will retrieve false results. Even if no hacker was involved, most developers buy the photographs that the AI system utilizes, since no one can physically take a million pictures. If is enough for one of these bought photographs to be poised or false for the algorithm to return with inaccurate results and predictions.
The Israeli company DeepKeep has developed an effective solution – an anti-virus to artificial intelligence. A system which detects anomalies, carries out examinations and even brings back reports regarding any vulnerabilities the AI engine might have.
The DeepKeep MLProtect platform provides tools to harden AI models and fortify them against attacks. By implementing defense methods such as AI firewall, detectors, pre-processing, and post-processing algorithms, the company creates a layer of protection which is much needed for the AI and ML systems of today. From the beginning of an AI system’s lifecycle all the way to complex pipelines, DeepKeep provides services to the HLS, automotive, fintech and insurtech industries, as well as many more. Furthermore, the first operational Adversarial AI system will soon be available to their customers.
The startup is a graduate of the INNOFENSE Innovation Center operated by iHLS in collaboration with IMoD. a unique acceleration program that removes entrance barriers to the technological ecosystem turning startups into mature, leading companies while connecting them with relevant investors, which is designed to strengthen the links between the civilian and defense markets via the collaborative development of the technologies, thus advancing and improving their integration in both markets.
For further information and inquiry, feel free to contact DeepKeep on their website at deepkeep.ai
Prepared to dive into the world of futuristic technology? Attend INNOTECH 2023, the international convention and exhibition for cyber, HLS and innovation at Expo, Tel Aviv, on March 29th-30th
Interested in sponsoring / a display booth at the 2023 INNOTECH exhibition? Click here for details!
