The U.S. federal government is urging artificial intelligence developers to prioritize security, cautioning that machine learning code is particularly challenging and expensive to modify after release. As part of its continuous push to encourage matching design and development programs with security from the outset, the Cybersecurity and Infrastructure Security Agency suggested that AI be secure by design in a blog post on Friday.
“Discussions of artificial intelligence frequently involve mysticism about how an AI system works. The truth is far more straightforward: AI is a category of software system. The government added that AI “must be secure by design, just like any software system.