Fastjson 1.x Unauthenticated RCE Exploited In-the-Wild (No Patch Available)

Technical Analysis

Tracked as CVE-2026-16723 with a CVSS score of 9.0, the vulnerability resides in Fastjson 1.x's deserialization/parsing logic, enabling attackers to send a specially crafted JSON payload to vulnerable Spring Boot endpoints and achieve remote code execution without authentication. This mirrors historical Fastjson autotype bypass RCE chains, where attackers abuse polymorphic deserialization to instantiate arbitrary classes and trigger gadget chains leading to code execution. No official patch exists yet, meaning affected deployments are currently unmitigated aside from workarounds like enabling safe mode or upgrading to Fastjson 2.x where feasible. Because Fastjson is a common dependency in Java-based backend services, including those powering RAG pipelines, tool-calling middleware, and orchestration layers for AI agents, successful exploitation could grant attackers full control of hosts running agent runtimes, exposing API keys, model credentials, and internal agent-to-agent communication channels.

Affected Systems

Applications using Fastjson version 1.x (all sub-versions prior to any forthcoming patch), particularly Spring Boot applications that deserialize untrusted JSON input via Fastjson's parse/parseObject methods without safe mode enabled.

Indicators of Compromise

• No specific hashes, IPs, or domains disclosed in source reporting; monitor for anomalous JSON payloads containing '@type' autotype markers targeting Spring Boot endpoints

Remediation Steps

1. 1

   Enable Fastjson Safe Mode

   Immediately enable Fastjson's SafeMode configuration to disable autotype deserialization, which blocks the primary exploitation vector for this class of vulnerability.

2. 2

   Migrate to Fastjson2

   Where feasible, migrate applications from Fastjson 1.x to the actively maintained Fastjson2 library, which has stronger deserialization protections.

3. 3

   Restrict Untrusted JSON Input

   Implement input validation and WAF rules to detect and block JSON payloads containing autotype/class-injection patterns before they reach application logic.

4. 4

   Network Segmentation and Least Privilege

   Run Java processes with minimal privileges and segment application servers from sensitive internal systems, including AI agent orchestration and credential stores, to limit blast radius if RCE occurs.

5. 5

   Monitor for Exploitation Attempts

   Deploy detection rules for suspicious JSON deserialization patterns and monitor Java process behavior for unexpected child process spawning or outbound connections.