gemini-3-flash-preview| input-price: 0.5 output-price: 3 max-context-length: 128_000 (cost: $0.01) · Anfrage: 2026-01-06T05:38:19.160855
Generated by rs-summarizer Version unbekannt
This material is best reviewed by Offensive Security Engineers, Vulnerability Researchers, and Mobile Product Security Architects. These specialists possess the requisite knowledge of heap memory management, protocol serialization (Protobuf), and low-level image parsing frameworks necessary to act on the technical findings presented.
Abstract:
This technical briefing deconstructs a sophisticated 0-click exploit chain targeting WhatsApp on iOS and Samsung devices, initially observed in-the-wild in 2025. The research identifies a critical two-stage attack: a logic flaw in WhatsApp’s "Linked Devices" protocol (CVE-2025-55177) used to bypass user interaction, and a subsequent memory corruption vulnerability in the DNG (Digital Negative) image parsing frameworks of iOS (CVE-2025-43300) and Samsung’s proprietary libimagecodec_qram. The presentation details the reconstruction of these exploits, the discovery of multiple 0-day vulnerabilities in Samsung’s image libraries during the investigation, and the persistent lack of robust hardware-backed mitigations (PAC, MTE) in third-party mobile applications.
Technical Breakdown and Key Takeaways:
RawCamera bundle. The vulnerability resides in the DNG (Digital Negative) decompression logic, specifically the Lossless JPEG unpacker.width * height * samples_per_pixel but utilizes a hardcoded loop limit of width * 2 rather than checking the actual num_components. This allows a malicious DNG with num_components = 1 to trigger an overflow.pair-data-operation-request (Protobuf type 16), which is white-listed for processing even from non-contacts.pair-device protocol message, forcing the victim's device to fetch a malicious DNG from an attacker-controlled URL.libimagecodec_qram, for DNG parsing. While initially appearing to be a simple integer overflow patch, the vulnerability was being exploited in-the-wild to target Samsung users.QRAM_DNG_Stream_get_data. A 32-bit integer overflow in the offset tracker results in a negative memory copy when processing files larger than 2GB.fix_bad_pixels_constant DNG OpCode. The research suggests that Samsung's image parsing surface remains highly vulnerable due to "subtle" patches and high code complexity.IP Service (used for AI-related image analysis) periodically scans stored images. This service automatically parses DNG files, providing a persistent background execution vector for compromised files.