Google is launching a fake call detection feature in its Phone app that flags when scammers spoof your contacts' numbers using AI voice cloning. The move comes as Americans lost over $893 million to AI-powered scams in 2025, according to FBI data. The feature automatically alerts users mid-call when it detects suspicious activity, giving them a chance to hang up before falling victim to increasingly sophisticated impersonation attacks targeting families and elderly victims.
Google just deployed a new weapon against the surge in AI-powered phone scams. The company's Phone app now includes fake call detection that warns users when they're likely talking to a scammer impersonating someone in their contacts - even when the caller ID looks legitimate.
The timing couldn't be more critical. Scammers are exploiting AI voice cloning technology to pull off increasingly convincing impersonation attacks, and the numbers tell a grim story. According to FBI Internet Crime Complaint Center data, Americans lost more than $893 million to AI-enabled scams in 2025 alone. That figure represents a sharp jump from previous years as voice cloning tools have become cheaper and more accessible.
Here's how the scam typically works. Fraudsters use number spoofing to make it appear they're calling from a contact already saved in your phone - maybe your mom, your boss, or your bank. Then they deploy AI voice synthesis to mimic that person's speech patterns and tone. The result is disturbingly convincing, especially in high-pressure scenarios where scammers claim there's an emergency or urgent financial need.
Google's solution runs entirely on-device, analyzing call patterns and behaviors that suggest fraudulent activity. When the system detects red flags - like a mismatch between the caller's supposed identity and suspicious requests for money or personal information - it displays a warning screen. Users can then decide whether to continue the conversation or hang up immediately.
The feature joins Google's existing arsenal of scam-fighting tools, including real-time scam detection that listens for common fraud patterns during calls. That earlier feature, announced at I/O 2024, used on-device machine learning to identify phrases and tactics typical of phone scams. The new fake call detection goes further by specifically addressing the AI impersonation threat.
What makes this particularly important is how these scams prey on trust. Unlike robocalls selling extended warranties, AI voice cloning attacks exploit emotional connections. Scammers impersonate grandchildren calling their grandparents, or bosses requesting urgent wire transfers from employees. The FBI has documented cases where fraudsters cloned voices using just a few seconds of audio scraped from social media posts.
Google says the feature will start rolling out this week to Android users through an update to the Phone app. The company didn't specify which Android versions or devices will get access first, but given the on-device processing requirements, it'll likely need relatively recent hardware with sufficient AI capabilities.
The broader industry is scrambling to respond to AI-powered fraud. Voice cloning services that once cost thousands of dollars now run for less than $10 per month. Some tools can generate convincing voice replicas from just a three-second audio sample, according to CNN reporting on AI voice scam trends. That accessibility has democratized sophisticated fraud in dangerous ways.
Google's approach prioritizes privacy by processing everything locally rather than sending call audio to cloud servers. That's a critical design choice given the sensitive nature of phone conversations, but it also means the feature's effectiveness depends on the quality of on-device AI models.
The company hasn't shared specific metrics on detection accuracy or false positive rates. Those numbers matter - if the system flags too many legitimate calls as suspicious, users might start ignoring the warnings. If it misses sophisticated scams, people remain vulnerable.
For now, security experts recommend a multi-layered defense. Use Google's new detection tools, but also establish verbal passwords with family members for emergency situations. If someone calls asking for money, hang up and call them back at a known number. And think twice before posting videos of yourself speaking on public social media profiles where scammers can harvest voice samples.
Google's fake call detection represents a necessary evolution in consumer protection as AI makes scams harder to spot. But it's also a reminder of a darker trend - the same AI technology driving innovation is being weaponized for fraud at scale. The $893 million Americans lost last year is likely just the beginning unless detection tools can keep pace with increasingly sophisticated attack methods. For users, the message is clear: treat unexpected calls with skepticism, even when they sound like someone you know.
