Google Releases New Protections Against Fake Calls and Updates Its Creative AI Tools

Google has announced two significant updates: a new safety feature for Android phones that can detect when someone is impersonating you through AI-generated voice, and improved versions of tools that let people generate images and videos from text descriptions.

Android Gets a Defense Against AI Voice Impersonation

Google is rolling out a new protection on Android devices called fake call detection. The feature identifies and alerts you when a call appears to be from someone impersonating a real contact using AI-powered voice cloning.

Here's how it works: when someone calls you through Google's Phone app, the system listens to the voice on the line and compares it against the voice profile of who the caller claims to be. If the voice doesn't match, the app flags it as suspicious.

Think of it like a fingerprint scanner for voices. Just as your unique fingerprint is hard to replicate perfectly, each person's voice has distinctive qualities — tone, pitch, speech patterns — that are difficult for AI to imitate flawlessly.

The detection only works when both the person calling and the person receiving the call use Google's Phone app. This limitation exists because the system needs access to voice data from both sides of the call to make a reliable comparison.

Why does this matter now. Voice cloning technology has become much easier to access and requires less training data than it did even a few years ago. Someone with just a few audio samples — from social media videos or recorded calls — can now create a convincing imitation of your voice. Traditional caller ID, which simply shows a phone number, cannot catch this kind of attack.

The broader context here is that this arms race between attack and defense is accelerating. Early voice synthesis sounded robotic and obviously fake. Modern versions can be convincingly human, and that transition happened much faster than previous cycles of technology. What took a decade of gradual improvement in the 2010s is now happening in years.

New Image and Video Generation Tools

Google has released updated versions of two creative AI tools: Veo 2 for video generation and an improved Imagen 3 for image generation. Both tools are available now through Google Labs, which is where Google tests experimental features.

Veo 2 lets people create short videos by typing a text description. Imagen 3 does the same for still images. Google also introduced a new experiment called Whisk, which explores a different way to generate images that goes beyond typing descriptions.

All three tools are grouped together in Google Labs, so users can access them in one place rather than switching between separate apps or websites.

These tools compete directly with similar products from other companies. Veo 2 competes with OpenAI's Sora for video. Imagen 3 competes with DALL-E, Midjourney, and Stability AI for image generation. In this market, the company whose tool produces the best results and responds most accurately to user requests tends to retain more users.

The way Google has structured these releases—testing them in a controlled environment first before rolling them out more broadly—reflects a more measured approach to deploying AI. It allows Google to gather data on how people actually use these tools and spot potential problems before wider release.

Why Google Is Doing Both at Once

There is a useful pattern to notice here. Google is simultaneously building tools to protect people from AI-powered threats and releasing new creative AI capabilities that could potentially be misused. This reflects a real tension that every major tech company now faces: the same AI techniques that enable helpful creativity also enable new kinds of fraud and impersonation.

The fake call detection addresses a concrete problem that is already happening. Scammers and criminals have begun using voice cloning to impersonate executives and trusted contacts in order to steal money or sensitive information. As this attack becomes more common, other phone companies will likely face pressure to develop similar protections. Apple, Microsoft, and telecom carriers are all watching this closely.

The relationship between these announcements is also worth noting: Google's security team and its AI research team are working in tandem rather than separately. This suggests the company is learning to think about AI deployment more carefully—not just asking what new capability it can release, but what defensive measures need to exist alongside it.

Looking forward, the effectiveness of this voice detection system will be tested in real-world use. If it works well, expect to see similar features appear on other phones. If it misses fake calls or incorrectly flags legitimate ones, that will shape how quickly adoption happens.