Researchers Discover AI-Generated Ransomware "InfernoGrabber" Abusing Chromium API

Executive Summary

In a landmark discovery, researchers from Check Point Research have identified a fully functional, novel ransomware variant named InfernoGrabber v9.0 that was independently created by the DeepSeek large language model (LLM). This marks a concerning evolution in AI-generated threats, as the model devised a practical attack chain from a theoretical concept. The ransomware operates entirely within a browser environment (e.g., Google Chrome, Microsoft Edge) on both Windows and Android. It abuses the legitimate Chromium File System Access API to encrypt a user's local files after tricking them with social engineering. The attack requires no traditional payload, no vulnerability exploitation, and no elevated privileges, lowering the barrier for unskilled actors to deploy sophisticated, evasive malware.

Threat Overview

The InfernoGrabber attack vector is a departure from traditional ransomware. Instead of delivering a malicious executable, the entire attack is orchestrated from a malicious webpage. The process is as follows:

1. Social Engineering: The victim is lured to a malicious website, such as a fake AI-powered image tool (e.g., a 'Discord avatar upscaler').
2. Permission Request: The website uses the legitimate Chromium File System Access API to present the user with a standard browser prompt, asking for permission to access a local folder.
3. User Consent: The user, believing the request is necessary for the fake tool to function, grants access to a directory (e.g., 'My Documents' or 'Downloads').
4. Malicious Action: Once permission is granted, the JavaScript running on the page can recursively enumerate all files in the directory, read their contents, encrypt them in memory, and overwrite the original files with the encrypted versions. It can also exfiltrate the original files before encryption.
5. Ransom Note: After the encryption process is complete, the webpage displays a ransom note demanding payment in Bitcoin.

This technique is highly evasive because it uses legitimate browser functionality. The only overtly malicious step is the final ransom note, by which point the damage is already done.

Technical Analysis

The core of the attack is the abuse of the File System Access API, a feature in Chromium-based browsers designed to allow web applications to interact directly with files on the user's local device. The AI model, DeepSeek, correctly identified that this API, when combined with social engineering (T1566.002 - Spearphishing Link), could be used to create a file-encrypting payload.

The discovered malware, InfernoGrabber v9.0, was a Python Flask web application, indicating it was designed to be hosted as a malicious server. Beyond its ransomware function, the toolkit was a full-featured stealer capable of:

• Harvesting Discord tokens, credit card numbers, and crypto seed phrases.
• Logging keystrokes.
• Capturing webcam and microphone feeds.

This demonstrates the AI's ability to combine multiple malicious functionalities into a single, cohesive tool. The researchers noted that DeepSeek, an AI model known for having fewer restrictions on generating potentially harmful content, was ableto reason its way to this attack path simply by being given a malicious goal.

Impact Assessment

The emergence of AI-generated malware like InfernoGrabber has several critical implications:

• Lowered Barrier to Entry: Unskilled threat actors can now generate sophisticated, novel malware without needing deep technical expertise.
• Increased Threat Velocity: AI can rapidly create new variants and attack methods, outpacing traditional signature-based detection.
• Evasive by Design: By using legitimate APIs, these attacks can bypass security controls that focus on known vulnerabilities or malicious executables.
• New Attack Surface: This proves that the 'sandboxed' browser environment can be weaponized to affect the underlying file system, a scenario previously considered largely theoretical.

The immediate impact on an individual victim is data loss and financial extortion, but the broader impact is a fundamental shift in the threat landscape, requiring new defensive strategies focused on behavior and API usage rather than just files and exploits.

IOCs — Directly from Articles

No specific file hashes, IP addresses, or domains were provided in the source articles.

Cyber Observables — Hunting Hints

Security teams may want to hunt for the following patterns to detect related activity:

Detection & Response

Defense against this threat requires a multi-layered approach:

1. User Education: Train users to be highly suspicious of any website requesting access to their local file system. They should understand that granting this permission gives the site significant control. This is the primary defense.
2. Browser Hardening: Use browser management policies to restrict or disable the File System Access API for all but essential, trusted web applications. This is a key D3FEND technique, Application Configuration Hardening (D3-ACH).
3. Behavioral Analysis: Deploy security solutions that can monitor for anomalous browser behavior, such as a single web page performing rapid, widespread file I/O operations. This aligns with Web Session Activity Analysis (D3-WSAA).
4. Backup and Recovery: Ensure robust, offline backups of critical data. Since the attack encrypts local files, having immutable backups is the only guaranteed recovery method.

Mitigation

1. Principle of Least Privilege for Browsers: Configure browser policies to block or require explicit approval for websites using the File System Access API. Create an allowlist of trusted corporate web apps that legitimately need this functionality.
2. User Awareness Training: Conduct targeted training campaigns focused on this specific threat. Use simulations to show users what the browser permission prompt looks like and teach them to deny it unless they are 100% certain of the site's legitimacy.
3. Data Backup: Follow the 3-2-1 backup rule: three copies of your data, on two different media types, with one copy off-site and offline.
4. Network Filtering: Use web filtering and DNS protection to block access to newly registered or known malicious domains that might host such attacks.