Microsoft unveiled Security Copilot, a GPT-4-based AI assistant that's aimed at cybersecurity professionals.
The chatbot and assistant system resembles ChatGPT but is built on "a small set of vetted security sources," according to Axios.
The automated system, which is built for enterprises, helps security teams analyze data and streamline information about ongoing cyber attacks as well as new threats.
- For example, teams can use it to ask natural language questions about security incidents at their companies and receive summaries from the chat interface.
- The prompts and responses are automatically saved, and users can pin the results in a shared workspace. It also analyzes feed files, URLs, and code snippets.
- OpenAI's GPT-4 and a trained custom model power Security Copilot. The latter model, which wasn't trained on customer data, "deploys skills and queries" related to cybersecurity and "incorporates a growing set of security-specific skills," Microsoft said.
- The system draws from information provided by Microsoft's threat intelligence, which takes in 65 trillion signals a day.
