The following podcast contains advertising.

To access an ad-free version of the LawFair podcast,

become a material supporter of LawFair at patreon.com slash law fair.

That's patreon.com slash law fair.

Also, check out LawFair's other podcast offerings,

rational security, chatter, law fair no bull, and the aftermath.

And that on the whole, it is better to put AI security within the broader context of cyber security.

I'm Stephanie Pell, senior editor at LawFair,

and this is the LawFair podcast, May 3rd, 2023.

Risks associated with the rapid development and deployment of artificial intelligence

are getting the attention of lawmakers.

But one issue that may not be getting adequate attention by policy makers or by the AI research and cyber security communities

is the vulnerability of many AI-based systems to adversarial attack.

A new Stanford and Georgetown report, adversarial machine learning and cyber security,

risks, challenges, and legal implications, offers a stark reminder that security risks for AI-based systems are real

and recommends actions that developers and policy makers can take to address the issues.

I set down with two of the reports authors, Jim Dempsey, senior policy adviser for the program

on Geopolitics, Technology and Governance at the Stanford Cyber Policy Center,

and Jonathan Spring, cyber security specialist at the cyber security infrastructure security agency, or SISA.

We talked about how AI-based systems are vulnerable to attack,

the similarities and differences between vulnerabilities in AI-based systems and traditional software vulnerabilities,

and how some of the challenges and problems with AI security may be social as much as they are technological.

...