According to foreign media reports, researchers at North Carolina State University have demonstrated a new way to attack artificial intelligence computer vision systems, allowing them to control what artificial intelligence “sees”. Research shows that the new technology, called RisingAttacK, can effectively manipulate all the most widely used AI computer vision systems.
What is being discussed here is so-called “adversarial attacks,” where someone manipulates data fed into an AI system to control what the system sees or cannot see in the image. For example, someone might manipulate AI’s ability to detect traffic lights, pedestrians, or other vehicles – which can cause problems for autonomous vehicles. Alternatively, hackers may install code on X-ray machines, causing AI systems to make inaccurate diagnoses.