Tag: large language model

In an effort to enhance automated discovery techniques, Google has created a new framework known as Project Naptime that, according to the company, enables a large language model (LLM) to conduct vulnerability research. According to researchers Sergei Glazunov and Mark Brand of Google Project Zero, "the interaction between an AI agent and a target codebase is centered around the Naptime architecture." A collection of specialized tools that are meant to resemble the work flow of a human security researcher are given to the agent. The project got its name since it lets people "take regular naps" while helping to automate variant analysis and conduct vulnerability research. The fundamental goal of the technique is to use the progress made in code comprehension and general reasoning ...