Apple's upcoming iOS 17.4 introduces Trust Insights, a new on-device framework designed to combat social engineering scams in real-time. The system analyzes interaction patterns, timing, and sensor data to detect when a user might be coached into falling victim to fraud across calls, texts, and emails. This aims to provide a proactive defense against increasingly sophisticated scams.

Trust Insights operates primarily on the device, assessing risk levels from medium to high. If a scam is suspected, apps can implement warnings, delays, or additional verification steps. Critically, Apple states the framework does not inspect content in Photos, Messages, or Mail. Instead, it analyzes behavioral signals, discards underlying data immediately, and sends only an output value for assessment.

This new framework will initially cover five operational categories: payment, account updates, resource use, communication, and a fallback 'other' category. Developers are encouraged to report feedback and confirmed fraud cases to help refine the system. Users can disable Trust Insights, though a cooldown period is built-in to prevent immediate disabling by a scammer.

The Trust Insights framework represents Apple's evolving approach to user security, moving beyond reactive measures to a more predictive, behavior-based defense against evolving fraud tactics.