These CVE IDs are still marked RESERVED at MITRE — no official description, no CVSS, no NVD record — yet the world is already talking about them. Mallory tracks the chatter so you see the risk before the paperwork catches up.
1,425 reserved CVEs with public mentions, ranked by all-time mention count.
Page 13 of 57
CVE-2024-6992 is a high-severity out-of-bounds memory-access vulnerability in ANGLE, the graphics-translation component used by Google Chrome and Chromium. In affected versions before Chrome 127.0.6533.72, a remote attacker could trigger heap corruption through a crafted HTML page.
CVE-2024-6992First seen Jul 30, 2026
CVE-2023-42219 is an Exim vulnerability associated with DNS resolution handling. It was reported as a limited information-disclosure issue when Exim relies on an untrusted DNS resolver. Available information does not identify the vulnerable function or disclose further technical exploitation mechanics. Some reporting inconsistently identifies this issue as CVE-2023-42119; the advisory mapping associates CVE-2023-42219 with Exim bug 3033.
CVE-2023-42219First seen Apr 14, 2026
First seen Aug 2, 2026
First seen Aug 26, 2026
CVE-2025-33143 is an uncontrolled recursion vulnerability in IBM Db2 for Linux, UNIX, and Windows, including Db2 Connect Server. An authenticated user can submit a specially crafted SQL statement that triggers uncontrolled recursive processing, resulting in denial of service. IBM Tivoli Business Service Manager versions 6.2.0.0 through 6.2.0.6 are affected where they include the vulnerable Db2 JDBC driver in the XMLToolkit component.
CVE-2025-33143First seen Aug 22, 2026
CVE-2025-6243 is a credential-management vulnerability in RUCKUS Network Director (RND). The platform contains a built-in user account, sshuser, that has root privileges, and both the public and private SSH keys for that account are stored in the account's home directory. Possession of the private key enables SSH authentication as the built-in privileged user. Because the account is pre-existing and highly privileged, the flaw effectively exposes a built-in administrative access path that can be used to obtain shell access on the affected RND server.
CVE-2025-6243First seen Jan 17, 2026
First seen Feb 4, 2026
First seen Jul 31, 2026
CVE-2025-8094 is a high-severity improper handling of permissions vulnerability in the GitLab Community Edition and Enterprise Edition project API. Under certain conditions, authenticated users with maintainer privileges could manipulate shared infrastructure resources beyond their intended access level through the project API. The flaw stems from insufficient permission enforcement in API operations governing access to shared CI/CD-related infrastructure resources, allowing a maintainer to act outside the intended authorization boundary. GitLab addressed the issue by refining permission checks within the project API to enforce proper access controls.
CVE-2025-8094First seen Jun 12, 2026
CVE-2026-31022First seen Apr 16, 2026
CVE-2026-39212 is a stack-based buffer overflow vulnerability in FFmpeg affecting ffmpeg_opt.c. The issue is described as a regression introduced in July 2025 and impacts FFmpeg command-line parsing or option-processing logic implemented in that source file. Successful exploitation occurs when attacker-controlled input reaches the vulnerable stack-resident buffer handling path, causing memory corruption through out-of-bounds writes on the stack. As a memory corruption flaw in a widely deployed media-processing framework, the vulnerability can lead to process instability and potentially more serious compromise depending on the surrounding execution context and exploitability of the overwritten stack data.
CVE-2026-39212First seen Jun 6, 2026
CVE-2026-39214 is a stack buffer overflow vulnerability in FFmpeg's Service Description Table (SDT) implementation. The flaw was reportedly introduced in 2003 and remained latent for many years. Based on the available information, the issue arises from unsafe handling of attacker-controlled input in SDT parsing logic, resulting in data being written past the bounds of a stack-allocated buffer. As a memory corruption flaw in media parsing code, successful exploitation could occur when FFmpeg processes a crafted media stream or container carrying malicious SDT data.
CVE-2026-39214First seen Jun 6, 2026
CVE-2026-39213 is a heap buffer overflow vulnerability in FFmpeg’s yuv4mpegenc component. The issue was reported as having been introduced in 2023. Available information identifies the flaw class and affected component, but does not provide the specific vulnerable function, code path, or triggering input structure. Successful exploitation would involve crafted media-processing input reaching the yuv4mpegenc encoding path and causing an out-of-bounds write on heap-allocated memory.
CVE-2026-39213First seen Jun 6, 2026
CVE-2026-39215 is a heap buffer overflow vulnerability in FFmpeg in the function update_mb_info(). The flaw was reportedly introduced in 2012. The available information identifies it as one of a set of FFmpeg memory-corruption issues affecting media-processing code paths. A crafted media input processed by the vulnerable code can cause out-of-bounds writes on the heap, leading to memory corruption and potentially destabilizing the process or enabling further exploitation, depending on allocator behavior and surrounding memory layout. Specific details about the exact source file, triggering format, and the precise bounds-checking failure are not currently available from the provided information.
CVE-2026-39215First seen Jun 6, 2026
CVE-2026-39211 is an integer overflow vulnerability in FFmpeg's swscale component. The issue was reportedly introduced in 2010. The available information identifies the flaw class and affected subsystem, but does not provide the specific vulnerable function, code path, trigger condition, or memory-safety consequences beyond the presence of an integer overflow. swscale is used for image scaling and pixel format conversion, so malformed or attacker-controlled media inputs that reach this processing path could potentially trigger incorrect size or arithmetic calculations during scaling operations.
CVE-2026-39211First seen Jun 6, 2026
CVE-2026-39217 is a heap buffer overflow vulnerability in FFmpeg's VP9 decoder. The issue is described as a regression introduced in March 2025. Successful exploitation occurs when the decoder processes attacker-controlled VP9 media data and performs an out-of-bounds write on heap-allocated memory. As a memory corruption flaw in a media parsing and decoding path, the vulnerability can be triggered during handling of crafted VP9 content by applications or services that rely on FFmpeg for decoding.
CVE-2026-39217First seen Jun 6, 2026
CVE-2026-39216 is a heap buffer overflow vulnerability in FFmpeg affecting code in img2enc.c. The flaw was reportedly introduced in 2012. The available information identifies the bug class and affected source file, but does not provide further technical detail about the specific vulnerable function, trigger condition, or parsing pathway. Successful exploitation would involve causing FFmpeg to perform an out-of-bounds write on heap-allocated memory while processing attacker-controlled input that reaches the vulnerable img2enc.c code path.
CVE-2026-39216First seen Jun 6, 2026
CVE-2026-33224 affects Bisheng and is described in the provided reporting as an authenticated remote command execution vulnerability in Bisheng's MCP tool/server configuration functionality. The issue is tied to unsafe handling of Model Context Protocol (MCP) STDIO-based server configuration, where attacker-controlled command and argument values can be introduced into local process execution without adequate sanitization or restriction. In the reported exploitation pattern, a user able to access Bisheng's MCP configuration workflow can register or modify an MCP tool/server definition so that Bisheng launches an attacker-specified operating system command on the host running the application. The reporting places this issue in a broader family of MCP STDIO command-injection flaws propagated through unsafe SDK and application design. It also notes that Bisheng's open user registration lowers the barrier to obtaining the authenticated access needed for exploitation.
CVE-2026-33224First seen Apr 20, 2026
CVE-2021-35406 is a command injection vulnerability in the Prolink PRC2402M router, tested on firmware version 20190909. Based on the provided supporting content, the flaw is in the /cgi-bin/qos.cgi CGI handler. When the page parameter is set to qos, the qos_settings routine processes attacker-controlled parameters including qos_dat and qos_bandwidth, uses sprintf to embed them into shell command strings, and executes the resulting command via do_system, a wrapper around system, without sufficient input validation or sanitization. This allows arbitrary shell metacharacters or command substitution to be injected through HTTP POST parameters. The provided proof of concept uses a crafted POST request to /cgi-bin/qos.cgi with page=qos and a malicious qos_dat value such as $(echo gg>/tmp/gg) to demonstrate code execution.
CVE-2021-35406First seen Apr 11, 2026
CVE-2024-35347 is a vulnerability in AMD CPUs, specifically affecting the microcode signature verification process. The flaw allows systems to accept microcode updates that lack the critical microcode signing fix, leaving them exposed to potential exploitation. The vulnerability primarily impacts AMD Family 19h CPUs and is documented in AMD security bulletin AMD-SB-7033. The root cause is insufficient enforcement of microcode signature verification, which can only be fully remediated by applying a BIOS update that delivers the updated microcode with the signing fix. Without this update, systems remain vulnerable even if they receive microcode updates from the Linux firmware repository, as these do not address the underlying signature verification issue.
CVE-2024-35347First seen Mar 18, 2026
First seen Mar 18, 2026
First seen Mar 18, 2026
CVE-2025-29617 is a critical vulnerability in the Piciorgros TMO-100 data modem for TETRA radio networks. The device exposes an unauthenticated TFTP service on both LAN and TETRA data networks, allowing any network-adjacent attacker to read and modify the modem's configuration. The configuration file contains sensitive information such as PPP credentials and network settings. Attackers can leverage this to alter port forwarding, disrupt modem operation, or facilitate further attacks. The issue affects software versions below 4.20, where TFTP access is unrestricted and unauthenticated.
CVE-2025-29617First seen Mar 18, 2026
CVE-2026-129820 concerns validation of incoming configuration hashes in the Perl Config::General module, distributed as libconfig-general-perl on Debian and perl-Config-General on Fedora. Upstream version 2.68 includes a fix. The precise validation defect, vulnerable function, and affected version range are not specified. The vulnerability is assessed as potentially affecting confidentiality, integrity, and availability.
CVE-2026-129820First seen Nov 17, 2026
CVE-2026-129821 is an input-validation vulnerability involving incoming configuration hashes in Perl Config::General, packaged as libconfig-general-perl on Debian and perl-Config-General on Fedora. Upstream version 2.68 includes a fix. The specific vulnerable function, validation failure, and exploitation mechanism are not documented in the available information.
CVE-2026-129821First seen Nov 17, 2026