How malevolent machine learning could derail AI
While artificial intelligence (AI) has the potential to revolutionize the technologies that people and organizations rely on in virtually any field, AI security expert Dawn Song warns that threat actors can cause great harm by abusing AI solutions through what is known as “adversarial machine learning.”
As scientists develop increasingly sophisticated AI solutions, criminals are looking for ways to manipulate those technologies for nefarious purposes. For instance, attackers can feed machine learning algorithms data that causes benign solutions to reveal sensitive information or to behave in malicious ways not intended by the developers. According to Song, adversarial machine learning is a “big problem” that should be addressed through a collaborative effort by researchers.