OpenAI’s GPT-4 Can Autonomously Exploit 87% of One-Day Vulnerabilities, Study Finds

Researchers from the University of Illinois Urbana-Champaign found that OpenAI’s GPT-4 is able to exploit 87% of a list of vulnerabilities when provided with their NIST descriptions.

ai artificial intelligence can cyber security descriptions developer exploit found gpt gpt-4 illinois list nist openai open source python researchers security study university vulnerabilities