Google introduced four new features; From AI content detection to fraud prevention

In a series of recent updates, Google has expanded the tool for detecting content produced by artificial intelligence, launched the Gemini 4 Argon flagship model, developed the possibility of running personalized applications made with artificial intelligence in Google Play, and introduced new security layers to deal with fraudulent messages and calls on Android. This [...] written by Google introduced four new features; From artificial intelligence content recognition

These capabilities were announced on October 7-14 and cover different areas of the Google ecosystem; From production and validation of digital content to software development and protection of mobile users.

1. SynthID Detector; AI content detection with invisible watermark

On October 7, Google expanded public and global access to the SynthID Detector tool. This service allows users to check if there is an invisible SynthID watermark on their desired image, video or audio file.

This technology has been developed since 2023 and is designed to identify AI-generated content through markers that are embedded in the digital output, without being normally discernible to the human eye or ear.

Google says that it has watermarked more than 180 billion images and videos and the equivalent of 240,000 years of audio content with this technology. The initial version of SynthID Detector was made available to media professionals last year, and now its access has been extended to the general public.

Images generated by artificial intelligence

SynthID watermarked videos

Audio and music files with this symbol

Google has announced that the tool will review content generated with its own models and some technology partners, including OpenAI, Nvidia and Kakao, and support for Apple is also announced for the future. However, the important thing is that SynthID Detector is not a general detector for all AI content; This tool relies on identifying compatible watermarks, and the absence of a mark does not necessarily prove that the content was not created by artificial intelligence.

The service is available through the SynthID Detector website, and Google has announced global availability in English.

2. Gemini 4 Argon; Google's new flagship model

Google unveiled Gemini 4 Argon on October 13; A model designed for complex reasoning, programming, scientific research, financial analysis, legal affairs, and cybersecurity.

According to Google, Argon performs better on multi-step tasks, provides more direct answers and, on long tasks, shows the progress of the activity more clearly to the user. This model is also integrated with the Google Antigravity agent-oriented development environment and enables the ability to move between conversations and execute tasks in the Gemini application.

The stated price for the API is a fee to use model based on the number of tokens and is not the same as Gemini's monthly subscription. Also, the cost comparison with competing models should be done according to the type of model, performance level, volume of use and pricing conditions; The claim of being cheaper in all scenarios cannot be concluded from these figures alone.

At the same time, Google has emphasized the security applications of the model. The company says its security teams have used Argon to probe systems, and through the Fairwind program, it provides cyber defenders with controlled access to a version of the model's capabilities.

3. Google Play made it possible to run personal applications made with artificial intelligence

On October 13, Google introduced a new feature for select users that makes it easier to run personal applications built in the Google AI Studio environment on Android devices.

With the help of artificial intelligence-based code generation tools, the user can create a program tailored to his needs and run it on the device without connecting the phone to the computer and without the need for a developer account. Early access has begun in select markets, including the US, for users 18 years and older.

Personal apps are kept separate from the public list of Google Play apps.

To run them, a standard Google account is sufficient and registration as a developer is not required.

Google plans to add the ability to share these apps with a limited group of friends and family members.

To publicly release an app, the developer must go through the usual release process and pass Google Play's security and quality checks.

This capability shortens the path of building small and personal programs; For example, a user can create a simple tool for managing expenses, daily planning or recording information. However, private execution of an application is different from public release, and passing security checks is still necessary for public release.

4. Android tackles scam messages and calls with the help of Gemini

On October 14, Google introduced a set of new security features for Google Messages and Android. The purpose of these features is to detect more sophisticated fraud and prevent the theft of sensitive information through messages, malicious links, fake calls and page sharing.

Google has announced that the current fraud detection systems in the Phone by Google and Google Messages applications flag more than five million suspicious calls and messages every month. This figure is related to Google's fraud detection systems and does not mean that all of these cases are blocked.

The most important new security features

Enhanced Protection Mode in Google Messages: This optional option allows for more advanced analysis of suspicious conversations with the help of Gemini cloud models. Google says the analysis takes place in the secure Private AI Compute environment and examines the context of conversations with anonymous senders to identify complex patterns of deception.

Check links with Gemini Nano: The lighter on-device model can use up-to-date web signals to evaluate anonymous links before opening them; A feature intended to counter newly created phishing websites.

Preventing secret call diversion: Android prevents apps from secretly activating call forwarding in the background and warns the user if they enter suspicious call forwarding codes. This action is designed to reduce the risk of theft of one-time passwords and accounts.

Hide sensitive information during page sharing: Password entry fields and payment pages in Chrome and Android WebView can be automatically hidden during live page sharing.

Verifying financial calls: In collaboration with participating banks, Android can verify whether an incoming call is actually initiated by a bank. If the call is not authenticated, in supported configurations the call is terminated and the user receives an alert. This feature is currently limited to supported banks, devices and markets.

Google also announced that fraud alerts in messaging notifications, which were first introduced to some Pixel phones, will be expanded to more devices in collaboration with other manufacturers. The release time and exact availability of each feature depends on the phone model, Android version and geographical region.

Four Google updates target different areas of everyday use of AI: SynthID Detector to investigate the provenance of some digital content, Gemini 4 Argon for complex and professional tasks, Google Play to run personal apps built with AI, and new Android features to combat fraud.

These possibilities are different in terms of scope and audience, but they all show that Google is expanding artificial intelligence from content creation tools to practical processes, software development and digital security. Their effectiveness in practice will depend on the quality of diagnosis, compatibility of devices, regional access and how users use it.