Google Expands AI Scam Defenses in India With New Tools

Google is deploying new AI-driven defenses for India to block a rising tide of digital frauds, banking on security innovations from Bengaluru that aim to secure Google Pay with an on-device call screen defense against scammers targeting Pixel 9 phone users, and trial its alerts interrupting synchronized screen-sharing cons across financial apps. It’s a significant improvement for a country that deals in massive scale on mobile — and yet there are some major blind spots still that jump out across Android and the wider app ecosystem.

What Google Is Doing to Counter Scams With On-Device AI

Its headline function is real-time scam spotting for Pixel 9 handsets. Running on Gemini Nano and working entirely on the device, the tool listens for established scam patterns in calls from “unknown” numbers and provides warnings to users mid-conversation. Google says the audio is not recorded or sent to its servers and that a beep in the call indicates when detection starts. It’s an opt-in feature and has launched in beta first for U.S. users in English, before the Indian launch.

Google also announced a pilot with Navi, Paytm and Google Pay to stop a fast-growing vector: screen-sharing scams. Scammers often convince people to share screens in order to record one-time passwords, PINs or account information. The pilot identifies problematic activity in real time and can alert or put a stop to sensitive actions when screen sharing is turned on — something Google first previewed at its developer conference and piloted in the United Kingdom.

The company says it’s working to put the call-protection feature on non-Pixel Android devices but hasn’t provided a timeline — and that should be noted because, like anything Android-related (and vice versa), when it comes to market share, Pixel makes up only a tiny percentage compared with its brand brethren.

Why India Is a Scam Hot Spot for Digital Financial Fraud

India’s rapid move to mobile-first finance has brought convenience and exposure, too. More than half of all bank fraud in 2024 was in digital transactions, leading to a reported loss of ₹5.2 billion from 13,516 cases (Prime Database via the Reserve Bank of India). Online scams cost around ₹70 billion in the first five months of 2025 alone, according to the Ministry of Home Affairs. Officials and researchers believe the underreporting is significant.

Unified Payments Interface adoption and low-friction onboarding have brought tens of millions of first-time internet users into the formal financial system. That scale, along with linguistic diversity and heavy smartphone reliance, has made social engineering the attack of choice — voice phishing, QR code baiting, “task” and “investment” schemes, and misuse of legitimate remote-access tools are among the recurring tropes.

Law enforcement has been responding more aggressively through the Indian Cyber Crime Coordination Centre and the national helpline 1930, but scammers’ changing scripts and infrastructure often outpace traditional takedown cycles.

Early Results and the Numbers From Google’s Protections

Google’s protective stack in India isn’t limited to calls. Play Protect has been tightening the tap on how many loan and utility apps requesting high-risk permissions (often abused for fraud) it lets through or via sideload; it tells me that this year, it blocked over 115 million installation attempts of these types. Within Google Pay, more than a million warnings are issued weekly for transfers suspected as being fraudulent by risk models.

For awareness, the DigiKavach campaign has so far covered over 250 million people; Google is working with the RBI to spread a list of authorized digital lending platforms backed by authentic non-banking financial companies. Google had previously announced a Safety Charter for India to bring AI-led defenses across products.

Where the Protections Are Still Inadequate and Risky

But even as on-device AI ups the ante, the ecosystem problem remains. Fraudulent or deceptive apps still slip through store reviews, especially cloned investment platforms and predatory lending apps that mimic legitimate services. Some go away only after police or researchers discover them — proof that reactive moderation can lag actual harm.

Side channels are another gap. APKs are often shared via messaging apps and social networks, so scammers can avoid the Play Store altogether. Play Protect scans can help, but victims who turn off protections or give broad permissions are still vulnerable. And call warnings currently apply only to numbers you don’t recognize, even if scammers commonly call from what appear to be known contacts (they sometimes spoof such identities, too), sometimes even after taking the time to earn that trust.

Coverage and context also matter. Any classifier will struggle with India’s linguistic diversity, code-mixing talk and quickly evolving scam scripts. There’s also the beep that comes with call analysis, which could act as a tip-off to criminals halfway through a scam. Finally, without a clear schedule for OEM-wide deployment, most Android users won’t see the call protection anytime soon, and many of the bad actors exploit VoIP channels or in-app calls that are harder to police.

What Banks and Regulators in India Will Want to See Next

Indian banks and regulators are calling for better default protections in high-risk flows: tighter overlay and screen-capture controls held by financial apps, tamper-resistant device attestation and clearer friction on abnormal transfers. Other coordinated and quick takedowns of malicious apps, accounts, and mule networks are priorities for the RBI, NPCI, and the Indian Cyber Crime Coordination Centre, with regular reporting also to CERT-In.

Transparency would help too. Banks say they want more granular numbers on false positives, language support and behavioral triggers to calibrate their own risk engines and not inconvenience legitimate users.

What to Watch Next as Google Rolls Out Anti-Scam Tools

What to watch for in the next few months: concrete plans to expand call-based scam identification beyond Pixel, more of a roll-out on screen-sharing protection across financial apps and signals that flagged attempts at fraud are actually happening less. It will be crucial to partner with carriers on network-level intelligence and wider language coverage.

For users, the basic playbook is straightforward:

• Don’t share screens during a financial transaction.
• Don’t install apps from links you receive from friends or unknown numbers.
• Check if potential lenders are on RBI’s authorized list of institutions, and call 1930 at the first hint of fraud.

Google’s new weapons are welcome — and overdue — but India’s fight against fraud will ultimately be won by using defenses in-depth, nimbler response times, and more “pain” for the channels scammers use most.