Apple iOS 27 Introduces AI Scam Detection for Third-Party Apps

Apple is preparing to introduce a new security feature in iOS 27 that will help apps detect potential scams in real time by identifying signs of social engineering attacks. According to Apple-focused publication 9to5Mac, the company has developed a new Trust Insights framework that enables developers to warn users when they may be manipulated into risky actions across phone calls, text messages, emails, and other communication channels.

Apple says these types of scams are difficult to identify because victims often carry out the requested actions themselves using legitimate authentication. To address this challenge, the Trust Insights framework uses privacy-focused machine learning to recognize behavioral patterns that may indicate a person is being guided into a fraudulent situation.

The company explained in a developer presentation that the framework relies on on-device AI to analyze interaction patterns, timing, context, and basic sensor data while protecting user privacy. Developers can integrate the system into their applications, interpret its risk assessments, and create appropriate safeguards without accessing sensitive personal content.

When suspicious activity is detected, the framework can classify the threat as either medium or high risk. Depending on the severity, supported apps can display warning messages, introduce delays before completing actions, or request additional verification to help users avoid potential scams.

The new capability arrives as AI-powered fraud continues to grow, with cybercriminals increasingly using deepfakes and other advanced techniques to deceive victims. Security experts believe AI can also play a valuable role in preventing these attacks. Jake Moore, global cybersecurity advisor at ESET, described the feature as “an excellent and positive use of AI,” noting that many people fail to recognize the behavioral tactics used by scammers. He added that the technology could significantly improve how users respond to fraudulent attempts.

Moore also cautioned that Apple must ensure the system is accurate and easy to understand. He said preventing scam payments without reading users’ messages or emails is encouraging, but warned that excessive false positives could eventually cause people to ignore security alerts.

As AI-driven fraud becomes increasingly sophisticated, Apple hopes its new Trust Insights framework will give developers powerful tools to help protect users while maintaining privacy. If the technology performs as intended, it could become one of the most significant security additions in iOS 27.