About once a week, I receive a message from my parents, my aunt, or some relative with a screenshot of what is to me some obvious spam or scam. To them, there’s always a question mark. Could this be real? Is someone really giving away $1000 to every person who clicks that link? So they ask me before clicking on the link or answering the message.

So yes, I am a living, breathing spam and scam detection system for my loved ones. Don’t get me wrong; I’m really glad they’re smart enough to be suspicious and ask someone who knows instead of just giving in to what the hordes of online scammers are luring them to do. I’m also proud of teaching and showing them enough examples in the past that they now know when to question the validity of something fishy.

Not everyone has a techie to turn to and double-check if something is a scam or not.

The problem is that not everyone in the world has a Rita or a Rita-like go-to techie in their lives. Fake humbleness and illusions of grandeur aside, I know that’s a real issue with real consequences.

France has a daily TV show dedicated to handling scams, including phone and message ones; US-based podcasts are filled with horror stories of people falling for these kinds of attacks; someone I know had to spend weeks fixing a security breach in an extremely well-known bank because an employee had clicked the wrong link in the wrong email (that’s all the detail I can give); and so on. As Dave Burke said, $1 trillion was lost to fraud in one year in 2023.

google io android gemini nano scam detection during calls

Still, even if this was universally available on all Android phones, I don’t think it’d be enough. Yes, it’s awesome to have a local AI looking out for you during a phone call, but scams don’t just happen during calls. Every day, I get phishing ads on the web because Google Ads allows them to go live to millions of people; I see scams on social platforms; I receive suspicious emails in my non-Gmail inboxes; or I get phishing messages on WhatsApp from unknown contacts. My phone exposes me to dozens of potential frauds in a week, but I’m used to them. Not everyone is.

Plus, even I dread the day when I’d click a link because I mistook a lower-case “l” for an upper-case “I” or an “rn” for an “m.” And I still double-check the URL each time my Google login times out and asks me to log in again. “Is this really Google I’m logging in to?” is a question I’ve asked myself dozens if not hundreds of times.

Our phones expose us to dozens of potential frauds in a week, so why not detect those too?

That’s why I’d love for Gemini Nano to expand its scam-detection potential beyond just Google’s apps and services. A true, local AI model should be able to handle everything on my screen and either hide it for me or alert me to its fraudulent potential. That’s the kind of useful AI intervention I want to see because not everyone grew up in the cautious age of the Internet, and people use many more email, messaging, and social platforms than just Google’s.