rootpwn

Cybersecurity

CrowdStrike Enhances Falcon Data Security with On-Device AI‑Driven Sensitive Data Classification

CrowdStrike has added a new on‑device AI capability to its Falcon Data Security product, enabling real‑time classification of sensitive information using on‑chip language models. The update improves data protection for endpoint and cloud workloads without relying on external inference services. Security teams should review the new classification settings and update their data‑loss‑prevention policies accordingly.

What Happened

CrowdStrike announced a new feature in Falcon Data Security that uses language‑model inference running directly on endpoint hardware. The on‑device AI classifies files and data streams in real time, flagging personally identifiable information, financial data, or other regulated content.

Impact

Endpoints and cloud workloads now receive instant, privacy‑preserving data classification without sending data to the cloud. The feature supports Windows, macOS, Linux, and major cloud platforms, and is available in Falcon XDR and Falcon Secure Access.

What to Do

Security teams should:

  • Enable the new “Data Classification” policy in the Falcon console.
  • Review classification rules and adjust sensitivity thresholds for your organization’s data categories.
  • Integrate the classification output with existing DLP and SIEM systems.
  • Monitor for false positives and update the language‑model tuning parameters if needed.
  • Document the change in your security policy and inform stakeholders.

Next Steps

Test the feature in a staging environment and validate that classification results match your compliance requirements. CrowdStrike provides documentation and a knowledge base article to guide configuration.

By moving AI inference to the device, we eliminate latency and protect sensitive data at the source, says CrowdStrike.

CrowdStrike AI Data Classification Falcon Endpoint Security On-Device AI Language Models

← All news