Technology Record - Issue 28: Spring 2023

79 work associated with user-reported phishing emails without creating any rules. Armorblox removes malicious emails across all mailboxes with just one click. In addition, customers can apply forward-looking remediation actions that automatically protect against similar attacks in the future, freeing security teams up to focus on threats that require human review. Armorblox connects to Microsoft 365 over Graph APIs and can run in parallel with any Microsoft licenses and complement the native investments into Microsoft. Redundant investments into secure email gateways can be cut down on as organisations look to complete their digital transformation strategy and go all-cloud with Microsoft 365 and other security investments. Large language models also fundamentally change how data protection is done inside organisations. Today’s data loss prevention solutions are riddled with false positives that take valuable time and resources away from security teams. Armorblox combines the traditional approach of custom regular expressions and identifiers with natural language techniques to match a wide array of sensitive and custom policy identifiers while reducing the number of false positives by 10 times. Investments made into Microsoft are leveraged to dynamically encrypt emails based on configured policy actions using the SmartHost configuration in the outbound direction, without requiring it to be an inline gateway with its own encryption logic. Armorblox can also use language models to automatically tag and classify data and documents, feeding this information back into Azure Information Protection over application programming interfaces. I have no doubt that GPT and large language models will change cybersecurity forever and new use cases will continue to arise. The application possibilities are endless – policy creation copilots, summarisation of logs, detecting drift in configuration settings across the network. Organisations should look to adopt large language models and AI in their security stack if they are to stay ahead of bad actors who will also be quick to leverage these advanced AI tools at their disposal. The GPT chapter is just beginning and more will be written and spoken about regarding regulation, compliance, GPT-generated threats, and more. It is an exciting time to be a part of the cybersecurity sphere as we usher in a new era of AI. Anand Raghavan is co-founder and chief product officer at Armorblox AI and large language models can help businesses reduce the risk of financial fraud