A €4,000 scam sparked a startup that stops phone fraud before your mobile even rings
Tech.eu Cate Lawrence
A scam that cost Rita Barbosa’s grandmother €4,000 led to Guardian Labs. Its app spots fraud on the phone itself, sometimes before it even rings.
Based on reporting by Tech.eu, Cate Lawrence — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Rita Barbosa didn’t start Guardian Labs because she wanted another AI startup. She started it after a phone scam pushed her grandmother into buying four water filters for about €1,000 each, even though each one was worth around €20. The money loss was bad enough. The bigger shock was what happened next: her grandmother lost confidence in using her own phone and stopped answering calls altogether.
That reaction sent Barbosa looking for a way to stop scams before they reached a victim. A hackathon in Germany gave the idea its first real shape. She won first place there with the prototype that became Guardião, and the project has since pulled in support from Portuguese police, two engineering universities in Portugal, one of the country’s largest accelerators, and the President of the Republic.
Guardião works as a layer between incoming calls or texts and the person receiving them. The AI runs locally on the phone, not on a remote server, which is the whole point. For messages, it checks for spoofing, phishing links and signs of manipulation before a notification appears. For calls, Guardian Labs says it can analyze the content before the phone rings at all, looking for urgency, emotional pressure and requests for banking codes. If it thinks the contact is fraudulent, it blocks it and then shows the user a notification, plus a history of what was intercepted and why.
Barbosa says that local processing only became practical recently, once small enough AI models existed to do the job on-device. The company is using scam reports from Portuguese police to fine-tune the system, along with data gathered through its own reporting flow. That matters because scams are not one-size-fits-all. In Portugal, for example, Barbosa points to the familiar “Hi Mum, Hi Dad, I lost my phone” trick, plus another common message scam pretending to come from the national healthcare service and demanding €46 or €47.
The company is also pushing beyond individual phones. It is building an intelligence platform for banks, telecom operators and law enforcement so active scam campaigns can be flagged at the institutional level. That could help identify mule bank accounts before victims send money. Guardian Labs says it is finalizing its first international pilots with banks and telecom companies, and plans to make money through B2B deals while keeping Guardião free for individuals.
The startup has moved fast for a team of three. After the German hackathon in February, Barbosa says it spread quickly across Instagram, Facebook and WhatsApp, with almost 20,000 followers on Instagram alone. That kind of growth says less about hype than about how painfully ordinary the problem is. People don’t need a demo to understand phone fraud. They already know someone who’s been hit by it.
My take — AI-written commentary, not fact-checked reporting
This is the rare AI pitch that doesn’t smell like a demo chasing a term sheet. Local-first scam blocking is sensible, especially when the alternative is handing your calls to some cloud server and calling it safety. The real tell is that the best selling point here is not magic; it’s avoiding the oldest trick in tech, which is making users absorb the risk so a platform can keep the data.
Read more about this at: Tech.eu