Microsoft introduces GPT-4 based AI assistant for cybersecurity tasks

Microsoft introduces Security Copilot, a tool intended for cybersecurity professionals. The tool is largely based on GPT-4 and combines data from different systems to monitor networks and assist employees.

Microsoft says that Security Copilot can be a useful tool for security professionals to quickly detect threats, respond to threats and gain a better understanding of the entire digital landscape of attacks and threats. The idea is that cybersecurity officers can better see what is happening, with Security Copilot learning from existing intelligence and identifying connections between threat activities.

Security Copilot accepts files, URLs, and parts of code for analysis.

Security Copilot is largely based on GPT-4 and Microsoft’s own security model. The latter model draws, among other things, from information from Microsoft Sentinel and Microsoft Defender, but acquisitions such as those from RiskIQ and Miburo should also help with the model. Microsoft says it uses 65 trillion daily signals collected as part of threat intelligence.

The tool looks like a simple window of a regular chatbot where employees can ask, among other things, what all security incidents within the company are, after which the bot displays them in summary. It is therefore primarily intended as a tool to assist employees rather than replace them.

Security Copilot is currently only available as a private preview and can only be used by a few companies for now. It is not yet known when the tool will become more widely available.