THRSP1060 Theater Tue 17 Nov 14:00 PST
Secure workplace AI without slowing innovation
Michael Jenkins
Summary
From Microsoft's AI summary
A ThreatLocker session argued that organisations should adopt AI with safeguards rather than ban it, warning that bans encourage shadow AI use. The speaker cited outages, Log4j, Kaseya and Follina as examples, said AI speeds vulnerability discovery and malware creation, and reported CVEs rising from roughly 5,000 per quarter in 2021 to over 35,000 in a recent quarter. Demonstrations showed guardrails bypassed by rewording prompts, and prompt injection was discussed. The speaker recommended zero trust measures including allow listing and ring fencing, and showed ThreatLocker controls.
Session description
Microsoft's catalogue description
AI tools have the power to transform organizational productivity while exponentially increasing risk. Uncover how workplace AI tools can expose sensitive data, bypass controls, and introduce new attack surfaces. Explore what you can do to mitigate these risks effectively without impeding innovation.