rootpwn

unknown

CVE-2026-100754

ChatGPT’s code‑signing validation on macOS and Windows can be bypassed by local, unprivileged users. This flaw allows an attacker to gain th

Overview

ChatGPT’s code‑signing validation on macOS and Windows can be bypassed by local, unprivileged users. This flaw allows an attacker to gain the agent’s trust, enabling prompt injection, reading of private chats, and access to TCC‑protected files. The vulnerability exposes sensitive data such as cookies, authentication tokens, and browsing sessions.

Description

ChatGPT's code-signing checks on macOS/Windows can be bypassed by (local, unprivileged) attackers, who then inherits the agent's trust to inject prompts, read private chats, access TCC-protected files and steal cookies/auth tokens/browsing sessions.

Impact

The compromise undermines confidentiality by exposing private conversations and authentication credentials. Integrity is threatened as attackers can inject malicious prompts. Availability is not directly impacted, but the loss of trust can erode user confidence. Defenders include system administrators, security teams, and end‑users of the affected platforms.

Remediation

1. Apply the latest security update from OpenAI that restores proper code‑signing validation. 2. Verify the integrity of the ChatGPT binaries using checksums or signed packages. 3. Enforce macOS TCC policies to restrict file access for the ChatGPT process. 4. Run the application in a sandboxed environment with the least privileges necessary. 5. Monitor for anomalous prompt injection or unauthorized file access.

Risk context

The CVE has an unknown severity and no EPSS score, indicating limited publicly available data on exploitation. However, the potential for credential theft and data leakage warrants timely patching and defensive hardening.

Affected products

  • ChatGPT macOS
  • ChatGPT Windows
  • OpenAI ChatGPT app
  • OpenAI ChatGPT web
  • macOS
  • Windows
  • OpenAI

Scores

Severity
unknown
CVSS v2
—
CVSS v3
—
CVSS v4
—
EPSS
—

code-signing macOS Windows ChatGPT TCC credential-theft sandboxing

← All CVEs

Related CVEs

unknown

CVE-2026-111862

Rank Math SEO PRO for WordPress is affected by a supply-chain compromise in a subset of version 3.0.123 packages downloaded or updated between October 15, 2026 at 20:02 UTC and October 16, 2026 at 07:12 UTC. Rank Math states that the affected plugin files contained malware that could compromise a website if the plugin was activated and directs sites that activated an affected package to treat the website as compromised. The malware may establish persistent access, communicate with rank1cdn[.]com, recreate hidden backdoors on page visits, and expose WordPress credentials, database credentials, API keys, and other secrets accessible to the web server. Rank Math identifies three malicious wp_options keys, hidden PHP backdoors, unauthorized administrator accounts, a fake WordPress core file, PHP-executable .cache files, and a system cron marker as incident indicators. Only packages obtained during the vendor-confirmed window are claimed as affected; version 3.0.123 alone is not sufficient to establish compromise. Rank Math removed the affected files and released version 3.0.124 as the clean replacement.

View analysis

unknown

CVE-2026-109720

A JIT miscompilation flaw in Mozilla Firefox’s WebAssembly engine can cause incorrect code execution when processing WebAssembly modules. The issue is limited to Firefox versions prior to 158 and the ESR 153.5 release. It was patched in those versions.

View analysis

unknown

CVE-2026-109721

A flaw in Firefox’s JavaScript JIT engine allows incorrect boundary checks. The issue could lead to memory corruption or crashes. It was addressed in Firefox 158 and ESR 153.5.

View analysis

unknown

CVE-2026-12980

The Post Snippets WordPress plugin (up to 4.2.4) fails to escape variable values in snippets, enabling Contributor-level users to inject arbitrary JavaScript that runs when content is viewed. This flaw allows attackers to execute malicious scripts in the context of site visitors. The vulnerability is a classic XSS flaw that can be exploited by anyone with Contributor or higher privileges.

View analysis