CVE-2026-23479: AI Discovers 2-Year-Old Redis RCE Vulnerability

Redis has patched a critical use-after-free vulnerability that allows authenticated attackers to execute arbitrary operating system commands on the database server. The flaw, tracked as CVE-2026-23479, was discovered by an autonomous AI security tool after remaining hidden in Redis code for more than two years. Consequently, organizations running unpatched Redis instances face significant remote code execution risk, especially in cloud environments where many deployments lack authentication.

What Happened: AI Discovers 2-Year-Old Redis RCE Vulnerability

On May 5, 2026, Redis released security updates addressing five vulnerabilities, the most severe of which is CVE-2026-23479. Furthermore, the flaw was introduced in Redis 7.2.0 and persisted through every stable branch until the recent patches. An autonomous AI bug-hunting tool built by Team Xint Code identified the vulnerability during analysis of the Redis codebase.

The issue exists in the unblockClientOnKey() function within src/blocked.c. When a key event wakes a blocked command, the function dispatches the queued command through processCommandAndResetClient(). However, this function can free the client as a side effect. The caller ignores the return value and continues reading the freed structure, resulting in a use-after-free condition that an attacker can chain into full remote code execution.

Technical Details of the CVE-2026-23479 Exploit Chain

The exploit chain operates in three distinct stages. First, a one-line Lua script leaks a heap pointer. Second, the attacker manipulates client memory limits, parks a bloated client on a Redis stream, drops the limits, and wakes the client. Redis frees the blocked client mid-call, and a pipelined SET command immediately reclaims the freed slot with a fake client structure. Third, Redis memory accounting performs an out-of-bounds decrement using attacker-controlled fields, overwriting a function pointer in the Global Offset Table to redirect strcasecmp() to system(). The next command Redis parses then executes as a shell command.

Prerequisites for Successful Exploitation

• An authenticated Redis session with access to CONFIG SET, EVAL, stream commands, and basic SET/GET operations
• The default Redis user typically holds all required privileges
• Partial RELRO in the official Docker image leaves the GOT writable at runtime

Affected Redis Versions

Branch	Affected Versions	Fixed Version
7.2.x	7.2.0 to 7.2.13	7.2.14
7.4.x	7.4.0 to 7.4.8	7.4.9
8.2.x	8.2.0 to 8.2.5	8.2.6
8.4.x	8.4.0 to 8.4.2	8.4.3
8.6.x	8.6.0 to 8.6.2	8.6.3

Business and Operational Impact

The cloud deployment footprint of Redis magnifies the risk of CVE-2026-23479. Wiz analysis indicates Redis runs in a large majority of cloud environments, and most instances operate without a password. Although the exploit requires authentication, the default user in typical deployments already possesses every privilege the attack chain requires.

Key Impact Areas

• Data Breach: Attackers can exfiltrate database contents, session tokens, and cached secrets
• Infrastructure Compromise: OS-level command execution enables lateral movement and persistent access
• Service Disruption: Attackers can corrupt data, terminate processes, or degrade application performance
• Compliance Exposure: Unpatched RCE flaws may violate SOC 2, PCI-DSS, and HIPAA security requirements

Mitigation and Recommendations

Redis has released patched minor versions for all affected branches. Minor upgrades within a series are designed to be drop-in replacements. Moreover, Redis Cloud customers are already protected, as the managed service has applied fixes automatically.

Immediate Actions for Defenders

1. Upgrade to the latest patched minor version for your Redis branch: 7.2.14, 7.4.9, 8.2.6, 8.4.3, or 8.6.3
2. Identify all internet-exposed Redis instances and place them behind firewalls or VPNs
3. Enforce strong authentication and disable default unauthenticated access
4. Tighten ACLs so no single role combines @admin, CONFIG, and @scripting privileges
5. Disable Lua scripting with deny @scripting if Lua is not required for your workloads
6. Rotate broadly shared Redis credentials immediately

Additional Hardening Steps

If immediate patching is not possible, restrict network access to trusted sources. Enable protected-mode in Community Edition and Open Source deployments. Ensure TLS encryption is in use for all Redis connections. Monitor for anomalous CONFIG SET, EVAL, and stream command usage from unexpected clients.

Bottom line: CVE-2026-23479 is a severe, authenticated remote code execution vulnerability hidden in Redis for over two years. Patch immediately, restrict network exposure, and enforce least-privilege access controls to reduce attack surface.

Incident Summary

CVE ID / Incident:	CVE-2026-23479 (NVD)
Affected Systems:	Redis server 7.2.0 through 8.6.2 (OSS, CE, and Software)
Disclosure Date:	May 5, 2026
Patch Status:	Available in versions 7.2.14, 7.4.9, 8.2.6, 8.4.3, and 8.6.3
Severity:	High (CVSS 3.1: 8.8; CVSS 4.0: 7.7)
Attack Vector:	Network-based, authenticated
CWE:	CWE-416 (Use After Free)

References

1. Redis, “Security advisory: CVE-2026-23479,” Redis Blog, May 2026, https://redis.io/blog/security-advisory-cve202623479/, accessed June 3, 2026.
2. National Vulnerability Database, “CVE-2026-23479,” NIST, https://nvd.nist.gov/vuln/detail/CVE-2026-23479, accessed June 3, 2026.
3. ZeroDay.Cloud, “Redis CVE-2026-23479 Deep Dive,” Team Xint Code, https://www.zeroday.cloud/blog/redis-cve-2026-23479-deep-dive, accessed June 3, 2026.
4. The Hacker News, “Autonomous AI Tool Finds 2-Year-Old RCE Flaw in Redis,” June 3, 2026, https://thehackernews.com/2026/06/autonomous-ai-tool-finds-2-year-old-rce.html, accessed June 3, 2026.
5. Theori, “Announcing Xint Code,” https://theori.io/blog/announcing-xint-code, accessed June 3, 2026.
6. Calif, “Codex Discovered a Hidden HTTP/2 Bomb,” Calif Blog, https://blog.calif.io/p/codex-discovered-a-hidden-http2-bomb, accessed June 3, 2026.