There's a scene in The Matrix when Morpheus explains to Neo that the Matrix is a world governed by rules and that those rules can be manipulated or broken. In effect, what Morpheus was advocating was finding ways to carry out an action that the software didn't anticipate in order to generate an unexpected outcome. That's what anti-AI is about.
In a discussion with Symantec's VP for global operations for cyber security services, John Lionato, and Peter Sparkes, the company's senior director for cyber security services in the APJ region, the spectre of AI was raised. And Anti-AI was mentioned as a tool for combatting the AI tools used by attackers.
This is software that works to intentionally disrupt the algorithms that drive machine learning.
Lionato talked about an experiment that was conducted using AI to illustrate how it works. Two identical images of his face were prepared and an AI tool was asked to identify whether the images were human.
However, one of the images was edited very subtly by just changing a few pixels. And those few pixels that were changed resulted in Lionato's face being identified as a giraffe even though the changes were imperceptible to other people.
This is what anti-AI is about. Sparkes said that as bad guys depend on AI in order to bypass security systems, they can be disrupted by using anti-AI. From the defence perspective, which is where Symantec is focussed, it means creating software that can disrupt an attack without compromising the user experience. However, he also noted that the idea of using Anti-AI tools like this was emerging and that research is progressing.