The gap between Android and iPhone is narrowing, and that includes exciting new security and privacy updates. And these days, it isn’t always Apple that takes the lead…
The prize for the most interesting AI security feature of the year so far could well be going to Google. The Android developer has suddenly revealed an entirely unexpected new feature that could be an absolute game changer for millions of smartphone users in the US and beyond.
The newly revealed Android feature sees on-device AI used to monitor voice calls coming into a smartphone in real time, warning users to potential scams—also in real time. Examples given included requests from purported bank staff that real bank staff are unlikely to ever make, such as requesting security codes or personal information, or teeing up unexpected transactions.
Because the analysis takes place entirely on-device using Gemini’s cutdown Nano platform, customer privacy is maintained. This is clearly critical for any form of call monitoring.
The on-device/off-device delineation is set to become a major factor as Samsung, Apple and Google battle it out in the premium smartphone AI stakes, launching new features without pulling too much user information into the cloud and causing privacy concerns.
This particular AI feature will apparently focus on the typical language patterns, words and phrases, and behaviors associated with fraud. And clearly the more widely it is deployed, the better it will become—subject to some form of training data cycle that fits the on-device privacy model.
In its “State of the Call” report for 2024, voice call specialists Hiya reported that “threats to the security and trustworthiness of voice calls also remain as prevalent as ever–and have only grown worse over the past year. In the last 12 months, more than 14 percent of all calls continue to be unwanted, while the average financial loss reported by consumers who fall victim to fraud calls reached $2,257. Meanwhile, businesses continue to lose revenue and incur higher operational costs due to challenges reaching customers who are wary of answering unidentified calls or calls flagged with spam or fraud labels—negatively impacting their brand reputation.”
From more than 200 billion calls analyzed, Hiya found that while “unwanted calls are sometimes labeled as spam or fraud on consumers’ mobile devices… many of the calls that consumers deem unwanted are not labeled at all… 92 percent of consumers believe unidentified calls are fraudulent. Nearly half—46 percent—of such calls go unanswered. In the case of the other half of unidentified calls—those that consumers pick up—recipients typically only answer reluctantly, due to concerns it may be a call they can’t miss.”
It is not yet known when this new feature will be released, and when it does it’s likely to be limited to newer devices with more capable AI processing onboard.
We have seen this disparity with Samsung’s “hybrid AI,” which limits the features being made available to older devices with the full suite restricted to new flagships. Google is expected to see the same with its latest Pixel devices. And all eyes will be on Apple’s iPhone 16 in the fall, with the focus of its new AI releases being on how older devices get to play, as well as on the shiny new hardware.
Google I/O 2024 continues, with ever more AI announcements set to dominate.
This post was originally published on 3rd party site mentioned in the title of this site