Adversarial algorithms can systematically probe large language models like OpenAI’s GPT-4 for weaknesses that can make them misbehave.
Source: Wired – Threat Level
A New Trick Uses AI to Jailbreak AI Models—Including GPT-4
- Cyberthreats linked to foreign actors aim to hinder Election Day proceedings
- DocuSign’s API used to lure victims into e-signing fake invoices
- Schneider Electric ransomware crew demands $125k paid in baguettes
- How to Become a Chief Information Officer: CIO Cheat Sheet
- Research: Extending corporate life of laptops by just one year can reduce harmful emissions by 25%