CVE-2026-14898
MediumCVSS 6.5Exploitation Probability (EPSS)
Low risk31th percentile - higher than 31% of all known CVEs
Summary
The OpenAI Codex desktop app for macOS automatically fetched remote images from model responses, allowing an attacker to inject an indirect prompt into content processed by Codex. The attacker could craft an image URL containing sensitive data, which was sent to an attacker-controlled server without user interaction. This vulnerability enabled exfiltration of secrets, API keys, source code, and data from connected tools.
Risk Assessment
The risk for the organization is the potential exfiltration of confidential data, such as API keys, source code, and results from connected tools, without user awareness. An attacker can exploit this vulnerability to steal sensitive information processed in the Codex session.
Recommendation
It is recommended to immediately update the OpenAI Codex application to the latest version that fixes this vulnerability. Until the update, avoid processing untrusted content in Codex and restrict access to the application.
Related vulnerabilities
- CVE-2026-115252Critical
The Confection WordPress Cable plugin from 0.4.0 to 0.4.7 does not perform capability or nonce checks before saving its settings and writes one of them unmodified into a standalone PHP file that it then serves. This allows unauthenticated users to execute arbitrary PHP code on the server.
- CVE-2026-124585Critical
A vulnerability was found in Chengdu Feiyuxing Technology VW2100 and VEC10GP devices within the web administration interface. The add_expmac function in var/html/webserver mishandles the mac argument, resulting in OS command injection. The attack can be performed remotely and a public exploit is available.
- CVE-2026-123870Critical
The otterdog webapp evaluates a format-string built from the incoming GitHub webhook JSON body using Python's eval(). When GITHUB_WEBHOOK_SECRET is unset or empty (default None), the HMAC signature check is skipped entirely, allowing an unauthenticated network client to reach the eval() sink with attacker-controlled input and execute arbitrary Python on the host.
- CVE-2026-124580Critical
A vulnerability was found in Chengdu Feiyuxing Technology VW2100 and VEC10GP version 1.0 B-MB3W202-210825-r11775, affecting the handle_wechat_cn_wifi function in var/html/webserver within the Weixin Auth Endpoint component. Manipulation of the extend argument results in OS command injection. The attack can be executed remotely and the exploit has been made public.
- CVE-2026-124577Critical
A weakness has been identified in Intelbras SG 2404 PoE L2+ versions 1.00.70 and 1.00.72, affecting an unknown function in the /EXCU_SHELL file within the Web Management Interface component. Manipulation can lead to improper authentication. The attack can be launched remotely and the exploit has been made publicly available.
- CVE-2026-113364Critical
Buffer overflow in the data receiving process in NEC WebSAM WinShare versions 7.5.0 and earlier, 7.6.0, and 7.7.0 through 7.7.2 on Windows allows a remote unauthenticated attacker to execute arbitrary code with SYSTEM privileges via a crafted packet with a malformed data size value in the protocol header.
- CVE-2026-113363Critical
Improper input validation in the network communication processing in NEC WebSAM WinShare versions 7.2.0 through 7.5.0, 7.6.0, and 7.7.0 through 7.7.2 on Windows allows a remote unauthenticated attacker to execute arbitrary commands with SYSTEM privileges via a crafted network packet sent to the computer running WinShare.
- CVE-2026-124560Critical
A vulnerability was found in D-Link DIR-878 version 1.20B05 in the GetCAPTCHAsetting function of bin/prog.cgi (HNAP Handler component). Manipulating the uid argument results in a heap-based buffer overflow. The attack can be initiated remotely and the exploit has been made public.
- CVE-2026-119506Critical
The Frontend Admin by DynamiApps plugin (acf-frontend-form-element) through version 3.30.1 contains an incorrect privilege assignment vulnerability. The flaw allows an attacker to escalate privileges within the WordPress site.
- CVE-2026-6715Critical
IBM Concert 1.0 through 3.0.1.1 allows an attacker to supply crafted input that is incorporated into an operating system command without adequate validation or sanitization. Successful exploitation allows remote attackers to execute arbitrary commands on the underlying operating system with the privileges of the affected service.
Original NVD description (English source)
The OpenAI Codex desktop app for macOS rendered remote images from Markdown in model responses. An attacker who could place an indirect prompt injection in content processed by Codex, such as a connected-tool result or another untrusted source, could induce the model to construct a remote image URL containing sensitive data. The app automatically fetched that URL when rendering the response, sending the embedded data to an attacker-controlled server without a separate user click. Successful exploitation could exfiltrate secrets and other information accessible in the Codex session, including API keys, source code, and data returned by connected tools. No direct integrity or availability impact was demonstrated, and there is no known exploitation in the wild.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

