CVE-2026-12199: NLTK WordNet Browser Remote Shutdown Vulnerability (7.5 CVSS)

Technical summary

The vulnerability exists in the `nltk.app.wordnet_app` module, where the HTTP server accepts an unauthenticated GET request to the path `/SHUTDOWN%20THE%20SERVER`. Upon receipt, the server calls `os._exit(0)`, terminating the process immediately. The server binds to all interfaces (0.0.0.0) in default mode and implements no authentication or authorization checks on this endpoint. This results in an unauthenticated, network-accessible denial-of-service vector with low attack complexity—a simple HTTP request is sufficient. The vulnerability is classified under CWE-306 (Missing Authentication for Critical Function), reflecting the absence of access controls on a critical server operation.

Business impact

For organizations using NLTK's WordNet Browser for linguistic analysis, research, or NLP services, this vulnerability enables adversaries to interrupt service availability at will. If the WordNet Browser is part of a production NLP pipeline or research infrastructure, repeated or sustained attacks could disrupt workflows and downstream systems that depend on it. The impact is availability-focused; data confidentiality and integrity are not threatened. However, the ease of exploitation and the default-insecure listening configuration create a practical risk of opportunistic attacks or internal disruption by low-privilege actors.

Affected systems

The vulnerability affects NLTK versions up to and including 3.9.3. Any deployment running the WordNet Browser HTTP server in its default configuration—particularly those listening on all network interfaces—is vulnerable. This includes research institutions, NLP development environments, and any service integrating the WordNet Browser for linguistic lookup. The default configuration is the most dangerous; custom deployments with restricted listening interfaces or authentication proxies may reduce exposure.

Exploitability

Exploitation is straightforward and requires no privilege escalation, authentication, or user interaction. An attacker needs only network access to the server and the ability to issue an HTTP GET request. The attack can be executed manually via curl, a browser, or automated tools. No special knowledge of NLTK internals or complex attack chains is required. The CVSS:3.0 vector (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) reflects this reality: network-accessible, low complexity, no privileges needed, and high availability impact. The absence of KEV status suggests this has not yet been observed in the wild with active exploitation, but the ease of exploitation makes proactive patching important.

Remediation

Upgrade NLTK to the latest version beyond 3.9.3 as soon as patch availability is confirmed with the NLTK project. Verify the patch version against the official NLTK release notes and security advisories. In the interim, restrict network access to the WordNet Browser server by binding it to localhost (127.0.0.1) only, unless remote access is mandatory. If remote access is required, deploy the server behind an authentication proxy or firewall that enforces authentication before requests reach the application. Disable or remove the WordNet Browser entirely if it is not actively used.

Patch guidance

Monitor the NLTK GitHub repository and official releases for version 3.9.4 or later. When a patch is available, upgrade via pip: `pip install --upgrade nltk`. Verify the fix by confirming the version and testing that the `/SHUTDOWN%20THE%20SERVER` endpoint either returns an error, requires authentication, or no longer exists. If running in a containerized environment, rebuild images with the patched version and redeploy. For production systems, test patches in a staging environment first to ensure compatibility with downstream NLP pipelines.

Detection guidance

Monitor HTTP server logs for requests to `/SHUTDOWN%20THE%20SERVER` or URL-encoded variations. Implement network intrusion detection signatures for this specific path. Track unexpected restarts or crashes of the WordNet Browser process, especially correlated with incoming HTTP requests from external sources. Query process execution logs for anomalous `os._exit()` or `sys.exit()` calls from the NLTK process. In containerized deployments, monitor for unexpected container restarts. Consider adding WAF rules to block requests to this path if the server is exposed to the internet.

Why prioritize this

This vulnerability merits prompt attention due to its combination of high CVSS score (7.5), ease of exploitation, and default-insecure configuration. The attack requires no credentials and can be executed in milliseconds. While the impact is limited to availability (not data breach), the availability impact is total and immediate—the process terminates. Organizations relying on NLTK for research or production NLP services should prioritize patching within their standard cycle. The lack of KEV status does not diminish urgency; it may indicate the vulnerability is newly disclosed or not yet widely exploited, making proactive defense valuable.

Risk score, explained

The CVSS 3.0 score of 7.5 reflects a HIGH severity rating driven by attack vector (Network), complexity (Low), privileges required (None), and availability impact (High). Although confidentiality and integrity are not affected, the complete and immediate denial of service to the application justifies the score. In a production environment integrating NLTK, this translates to potential service downtime with minimal attacker effort. The score does not account for business context (e.g., mission-critical linguistic analysis), so organizations with high availability requirements should consider this a critical fix.

Frequently asked questions