
The TED AI Show: How AI is changing national security w/ Kathleen Fisher
TED Tech
00:00
Vulnerabilities in Large Language Models: Universal Suffix Attacks
This chapter examines the vulnerabilities of large language models, emphasizing malicious techniques like universal suffix attacks that can alter AI outputs. The discussion highlights the differing security approaches of AI developers, with a particular focus on Anthropic's measures to protect their models.
Play episode from 29:25
Transcript


