Description

Security researchers have identified three critical vulnerabilities in Authlib, a widely used library for implementing OAuth and OpenID Connect, with over 86 million monthly downloads. The flaws allow attackers to bypass authentication, forge tokens, and decrypt sensitive data. The most severe issue (CVE-2026-27962, CVSS 9.1) enables attackers to forge valid JWT tokens by injecting their own cryptographic keys. Another flaw (CVE-2026-28498, CVSS 8.2) allows authentication bypass through a fail-open validation of ID tokens. The third vulnerability (CVE-2026-28490, CVSS 8.3) exposes encrypted data via a padding oracle attack due to improper error handling. These issues affect all Authlib versions up to 1.6.8 and impact applications built on frameworks like Flask, Django, and FastAPI. The vulnerabilities arise from improper cryptographic implementation and insecure validation logic. Authlib’s handling of RSA1_5 encryption unintentionally creates a reliable oracle through distinct error responses, enabling attackers to decrypt sensitive data. Additionally, the library fails to enforce strict validation of token algorithms, accepting unrecognized values as valid. The most critical flaw occurs when applications rely on externally supplied keys during token verification, allowing attackers to manipulate trust boundaries and bypass authentication controls entirely. Organizations should immediately upgrade Authlib to version 1.6.9 or later to remediate these vulnerabilities. Developers must enforce strict token validation, ensuring that verification keys are never derived from untrusted token input. Security teams should disable deprecated algorithms like RSA1_5 and adopt modern, secure cryptographic standards. Additionally, organizations should audit authentication flows, implement robust error handling, and continuously monitor systems for abnormal authentication activity.