CrowdStrike has introduced a slate of new features to its Falcon platform aimed at tackling emerging cybersecurity risks from malicious AI models and data exfiltration, SiliconAngle reports. The update adds specialized capabilities to detect and block threats hidden within artificial intelligence systems, alongside improved visibility into data misuse across endpoints and cloud environments.A core addition is the AI Model Scanning feature, which analyzes large language models for embedded backdoors or malicious behaviors. These so-called “trojanized models” may function normally in most use cases but produce harmful output when triggered by specific inputs. The tool leverages CrowdStrike’s ExPRT.AI technology to rank identified threats by severity, helping security teams prioritize issues based on real-world attack intelligence.CrowdStrike has also rolled out an AI Security Dashboard to track shadow AI usage—instances where employees use unauthorized or unsecured AI tools. This dashboard helps organizations monitor how internal training data is accessed and potentially misused, strengthening guardrails around proprietary information in an AI-driven workplace.On the data protection front, the Falcon platform has been enhanced to detect risky file transfers, including obfuscated exfiltration methods like encrypted ZIP archives. New capabilities allow for real-time scanning of such files and flag attempts to upload modified sensitive data to generative AI tools. Additionally, Falcon Data Protection for Cloud now extends these controls to public cloud workloads, while a new SaaS Threat Services offering provides expert-led scans of third-party application risks.
You can skip this ad in 5 seconds