Adversarial algorithms can systematically probe large language models like OpenAI’s GPT-4 for weaknesses that can make them misbehave. Read the story <a href="https://www.wired.com/story/automated-ai-attack-gpt-4/">here</a>. 

Learn about your ad choices: <a href="https://dovetail.prx.org/ad-choices">dovetail.prx.org/ad-choices</a>

Adversarial algorithms can systematically probe large language models like OpenAI’s GPT-4 for weaknesses that can make them misbehave. Read the story here. Learn about your ad choices: dovetail.prx.org/ad-choices

Adversarial algorithms can systematically probe large language models like OpenAI’s GPT-4 for weaknesses that can make them misbehave. Read the story <a href="https://www.wired.com/story/automated-ai-attack-gpt-4/" rel="noopener noreferrer">here</a>. 

Learn about your ad choices: <a href="https://dovetail.prx.org/ad-choices" rel="noopener noreferrer">dovetail.prx.org/ad-choices</a>

A New Trick Uses AI to Jailbreak AI Models—Including GPT-4

Get in-depth coverage of current and future trends in technology, and how they are shaping business, entertainment, communications, science, politics, and society.

Technology

Get in-depth coverage of current and future trends in technology, and how they are shaping business, entertainment, communications, science, politics, and society.

Share A New Trick Uses AI to Jailbreak AI Models—Including GPT-4

Sign up to save your podcasts

A New Trick Uses AI to Jailbreak AI Models—Including GPT-4

A New Trick Uses AI to Jailbreak AI Models—Including GPT-4

More shows like Security, Spoken

Security Now (Audio)

SANS Internet Stormcenter Daily Cyber Security Podcast (Stormcast)

What's New

Science, Spoken

Business, Spoken

Smashing Security

The Daily

Defense in Depth

Citadel Dropouts: a Game of Thrones Podcast