CVE-2026-48793: Jellyfin Unauthenticated Argument Injection via Subtitle Conversion

Technical summary

CVE-2026-48793 is an argument injection vulnerability in Jellyfin's SubtitleEncoder.ConvertTextSubtitleToSrtInternal method (SubtitleEncoder.cs, line 382). The code constructs FFmpeg command-line arguments by interpolating a subtitle file path without calling EncodingUtils.NormalizePath(), which is the function responsible for safe path escaping. On Linux systems, where filenames can contain unescaped double-quote characters, an attacker can craft a subtitle filename that breaks out of the argument quoting context and inject arbitrary FFmpeg arguments. The vulnerable endpoint SubtitleController.GetSubtitle lacks an [Authorize] attribute, making it accessible without authentication. FFmpeg's extensive command-line interface then allows arbitrary file operations including write and information disclosure.

Business impact

A successful exploit grants an attacker the ability to read arbitrary files from the Jellyfin server (including configuration, private keys, or other sensitive data stored locally) and to write or overwrite files on the host system. For organizations running Jellyfin in production—particularly those hosting on shared infrastructure or with untrusted users who can contribute media files—this represents a severe operational and confidentiality risk. If the Jellyfin process runs with elevated privileges, the impact extends to the underlying system.

Affected systems

Jellyfin versions prior to 10.11.10 are vulnerable. The issue affects all operating systems where Jellyfin runs, but the injection mechanism is most reliably exploited on Linux due to its permissive filename handling. The vulnerability is reachable without authentication, making it relevant to any Jellyfin instance accessible over the network, regardless of internal access controls.

Exploitability

The attack is straightforward to execute. An attacker needs only the ability to place or upload a file into a Jellyfin media library directory (such as a shared NAS, Samba share, guest upload folder, or any directory Jellyfin monitors for media). They then trigger subtitle conversion by requesting the file via the unauthenticated GetSubtitle endpoint. No special tools, kernel exploits, or complex interaction is required. The lack of authentication and the direct network reachability of the vulnerable endpoint make this a practical attack vector.

Remediation

Immediately upgrade Jellyfin to version 10.11.10 or later. This release patches SubtitleEncoder to properly normalize subtitle file paths using EncodingUtils.NormalizePath() before passing them to FFmpeg. Additionally, restrict filesystem write permissions for the Jellyfin process to the minimum necessary directories, implement network-level access controls to limit who can reach Jellyfin's HTTP endpoints, and audit media library shares to remove or quarantine any untrusted or guest-accessible upload paths.

Patch guidance

Upgrade to Jellyfin 10.11.10 as soon as possible. For organizations unable to patch immediately, disable or remove guest upload functionality from media library paths and restrict network access to the Jellyfin server to trusted users only. Monitor Jellyfin logs for unusual subtitle conversion requests or FFmpeg argument errors, which may indicate exploitation attempts.

Detection guidance

Hunt for HTTP requests to SubtitleController.GetSubtitle endpoints that reference subtitle files with suspicious filenames containing double-quote characters or shell metacharacters. Monitor FFmpeg process execution on Jellyfin servers for unexpected arguments or suspicious file operations (reads from /etc, /root, ~/.ssh, or writes outside expected media directories). Check Jellyfin logs for conversion errors or failed FFmpeg invocations. On shared storage systems, audit media library directories for files with names designed to break argument quoting (e.g., containing quotes, backticks, or unusual characters).

Why prioritize this

This vulnerability merits immediate attention due to its combination of high impact (arbitrary file read and write), ease of exploitation (no authentication, simple file placement), and broad reachability (network-accessible, unauthenticated endpoint). Organizations running Jellyfin with network exposure should treat this as a critical priority. The vulnerability is not yet in the CISA KEV catalog, but its practical exploitability and the open-source nature of Jellyfin mean exploit code could emerge rapidly.

Risk score, explained

The CVSS 3.1 score of 8.8 (HIGH) reflects high impact (confidentiality, integrity, and availability all rated High), low attack complexity, network reachability, low privileges required, and no user interaction. The score accurately captures the severity: unauthenticated attackers can execute arbitrary code-like operations (via FFmpeg) with file system consequences. The main factor preventing a critical score is that successful exploitation still depends on the attacker's ability to place a file in the media library, though this is often feasible in shared or guest-accessible environments.

Frequently asked questions