The vulnerability identified as CVE-2026-28348 affects the lxml_html_clean library, a Python project providing HTML cleaning functionalities derived from lxml.html.clean. The core issue lies in the _has_sneaky_javascript() method, which attempts to detect and filter dangerous CSS constructs such as @import and expression() that could be used for malicious purposes. However, prior to version 0.4.4, this method improperly strips backslashes before performing keyword checks, allowing CSS Unicode escape sequences to evade detection. This bypass enables attackers to inject external CSS or execute cross-site scripting (XSS) attacks, particularly in older browsers that interpret these escape sequences. The vulnerability is categorized under CWE-116, indicating improper encoding or escaping of output. The CVSS v3.1 base score is 6.1, reflecting a medium severity level with network attack vector, low attack complexity, no privileges required, but requiring user interaction. The scope is changed, indicating that the vulnerability can affect components beyond the initially vulnerable library. Although no known exploits have been reported, the flaw poses a tangible risk to applications relying on vulnerable versions of lxml_html_clean for sanitizing HTML content. The issue was publicly disclosed and patched in version 0.4.4, which corrects the filtering logic to properly handle CSS Unicode escapes and prevent bypasses.

This vulnerability can lead to the injection of malicious CSS or JavaScript code via bypassing sanitization filters, resulting in cross-site scripting (XSS) attacks or unauthorized external CSS loading. For organizations, this can compromise the confidentiality and integrity of user data, enable session hijacking, defacement, or execution of arbitrary scripts in the context of trusted web applications. The impact is particularly significant for web applications that rely on lxml_html_clean for sanitizing user-generated HTML content before rendering it in browsers. Older browsers are more susceptible to exploitation due to their handling of CSS Unicode escapes. While availability is not directly affected, the reputational damage and potential data breaches from successful XSS attacks can be severe. Since the attack requires user interaction, phishing or social engineering could be used to lure victims. The vulnerability affects any organization using vulnerable versions of the library, especially those in sectors with high web application usage such as finance, e-commerce, healthcare, and government services.

Organizations should immediately upgrade lxml_html_clean to version 0.4.4 or later, where the vulnerability is patched. Review and audit all instances where lxml_html_clean is used for HTML sanitization to ensure no legacy versions remain in production. Implement additional defense-in-depth measures such as Content Security Policy (CSP) headers to restrict the execution of unauthorized scripts and loading of external CSS resources. Employ rigorous input validation and output encoding on all user-supplied content, especially in web applications accessible via older browsers. Conduct security testing focusing on CSS injection and XSS vectors to detect any residual weaknesses. Educate developers about the risks of improper encoding and escaping, and encourage the use of well-maintained libraries for sanitization. Monitor security advisories for updates or emerging exploits related to this vulnerability. If upgrading immediately is not feasible, consider applying temporary filters or disabling features that rely on vulnerable sanitization until patched.