CVE-2026-18564 in LibreOffice
Summary
by MITRE • 11/16/2026
LibreOffice can export documents to PDF protected by a password. With the PDF 2.0 option the file is encrypted with 256-bit AES, using a file encryption key and salts that are meant to be random. The key and the salts were drawn from a pseudo-random number generator that is not suitable for cryptography, so the encryption key of an exported PDF could be recovered without knowing the password. In fixed versions the key and the salts are drawn from the random source of the operating system.
If you want to get the best quality for vulnerability data then you always have to consider VulDB.
Analysis
by VulDB Data Team • 11/16/2026
The vulnerability in LibreOffice stems from a critical implementation flaw in its document export functionality when generating PDF files with 256-bit AES encryption under the PDF 2.0 standard. While the intention was to provide robust security by using file encryption keys and salts derived from random sources, the underlying mechanism relied on a pseudo-random number generator that lacked cryptographic strength. This means that although the output appeared secure due to the use of strong symmetric encryption algorithms like AES-256, the entropy source used to generate the necessary keying material was predictable or statistically biased. An attacker with access to multiple exported PDF files could potentially analyze the patterns in the generated keys and salts to reverse-engineer the state of the pseudo-random number generator. This allows for the recovery of the encryption key without knowledge of the user-provided password, effectively bypassing all confidentiality protections intended by the document owner.
From a technical perspective, this flaw represents a classic case where strong cryptographic primitives are undermined by weak random number generation. The security of AES-256 relies entirely on the secrecy and unpredictability of its keys. If an attacker can predict or reconstruct these keys through statistical analysis of multiple outputs from the flawed generator, the encryption becomes trivial to break. This is not a weakness in the AES algorithm itself but rather a failure in key management practices within the application layer. The salts, which are intended to ensure that identical plaintexts produce different ciphertexts and prevent pre-computed attacks like rainbow tables, were also derived from this insufficient source, further compounding the risk by allowing attackers to correlate encrypted documents or accelerate brute-force attempts if partial information is known.
The operational impact of this vulnerability is severe for any organization relying on LibreOffice to distribute sensitive confidential documents via PDF format with password protection. Users who believe their data is secure due to AES-256 encryption may have a false sense of security, as the actual barrier against unauthorized access is significantly weaker than advertised. Attackers could potentially decrypt protected reports, financial statements, or legal documents without needing to crack passwords through brute force or social engineering. This undermines trust in digital document workflows and exposes proprietary information to competitors, malicious actors, or public disclosure depending on how the files are shared. The risk is particularly acute when these encrypted PDFs are transmitted over networks or stored in cloud environments where they might be intercepted or accessed by unauthorized parties who possess sufficient samples of previously exported documents for analysis.
To mitigate this vulnerability and align with industry best practices, it is imperative to upgrade LibreOffice to a version that has addressed the random number generation issue. The fix involves replacing the application-level pseudo-random generator with calls to the operating system's cryptographically secure random source, such as /dev/urandom on Linux or BCryptGenRandom on Windows. This ensures that keys and salts are generated using high-entropy sources designed specifically for cryptographic purposes. Organizations should also implement a policy of regularly updating office suite software to patch known vulnerabilities in security-critical components like encryption modules. Additionally, users handling highly sensitive information should consider employing additional layers of protection or verifying the integrity and version of their document processing tools before exporting confidential data. This vulnerability is categorized under CWE-330: Use of Insufficiently Random Values, which highlights the danger of relying on non-cryptographic random number generators for security-sensitive operations. In terms of attack vectors, this relates to ATT&CK techniques involving exploitation of software vulnerabilities and potentially credential access if passwords are reused or weak, although the primary exploit here is technical decryption rather than password cracking.