Google's new feature aimed at alerting users to the possibility of scams has caused concern.

The tool uses artificial intelligence technology to listen to people's phone calls and determine whether they sound like a scam. If so, a window pops up warning users of a "possible scam".

The feature was unveiled at this week's I/O event, where Google announced new AI tools. Google did not say exactly when this feature will be available.

"INCREDIBLY DANGEROUS"

The company also did not specify what kind of conversations would trigger the AI's fraud detection. But the company said the feature is based on the recently launched Gemini Nano, a much smaller version of its AI designed to work on phones.

Google said that all listening and analysis of phone calls will be done on that phone, meaning that private conversations will not be sent to its servers. "All of this protection happens on the device, so your conversation stays private," it said in its announcement.

But security experts nonetheless suggested that intercepting phone calls in this way was "incredibly dangerous" and "scary".

"PAVES THE WAY FOR SCANNING CONVERSATIONS"

"Candy world" discovered beyond Neptune! "Candy world" discovered beyond Neptune!

Even if calls remain on the device, allowing AI to listen in on calls could lead to other problems, it said.

Meredith Whittaker, head of messaging app Signal, said: "This is incredibly dangerous. It paves the way for device-level and centralized scanning of user conversations."

Whittaker, who worked for 13 years at Google and helped organize internal protests against its policies, said the use of the technology could quickly become widespread.

Editor: David Goodman