Android uses AI to check Google Messages and calls for possible fraud attempts

Google has announced a new security function for Android devices that uses artificial intelligence (AI) to detect attempted fraud. The system monitors messages on Google Messages, as well as calls, and warns the user of potential scams in real time. The company promises that the function will only run on the device and that conversations will remain private. Initially, scam detection in Google Messages will be limited to English-speaking countries, but will be extended to other regions in the future.

This AI security function is an extension of the protective measures already introduced for Google Messages in October last year. This also involved protection against fraud attempts, but also warnings about potentially dangerous links. Many users have only recently noticed the Android System SafetyCore, a nude photo warning system for smartphones. In this case, too, Android monitors the messages and warns of potentially sensitive content.

AI trained to detect sophisticated and common fraud attempts

Now, however, the data company is emphasizing the use of AI to detect fraud attempts. According to its security blog, Google has sought support from financial institutions worldwide to understand the most sophisticated and common scams affecting their customers. The message exchange often starts out harmless, but after a while the victim is manipulated into handing over private data, transferring money or switching to other messenger apps.

The usual scam protection warns users of possible fraud attempts before the conversation begins and is based on the phone number. However, many scammers disguise these or use the telephone numbers of trustworthy companies if the victim is aware of them. This is why the exchange of messages in Google Messages will be continuously monitored in the future, so that Android will also issue warnings during the conversation if suspicious messages are detected.

Monitoring private conversations in this way raises alarm bells. However, Google promises that the AI security function will only be executed locally. The messages would remain on the Android device. If the user reports a detected scam attempt, only the details of the scammer and their last messages are transmitted to Google and the mobile phone provider.

Scam check also for calls

This also applies to fraud detection for calls, which Google introduced for some Pixel smartphones in the USA at the end of 2024. Android emits a warning tone and vibrates if the call is about payments, for example. According to Google, the calls are not recorded or transmitted here either. In contrast to scam detection in Google Messages, however, this function is not activated by default for calls; this is up to the user.

The new AI security function for Google Messages will initially be introduced in English in the USA, Canada, and the UK. An expansion to other countries is to follow shortly, but it is unclear when this will also take place in German. In this country, Telekom's SMS firewall, which the German mobile network operator has just announced for April, will probably come first. This is intended to protect recipients from malicious messages that could cause them to install malware or fall for scams. However, this is limited to traditional text messages.

(fds)