Just published.
The newest CVEs, straight from CVE.org and NVD.
-
CVE-2026-116104
A security flaw has been discovered in Sony Neural Network Libraries up to 1.39.0. This impacts the function func_set_nnabla_version_decorate of the file python/src/nnabla/utils/converter/utils.py of the component NNP Converter. Performing a manipulation of the argument function.type results in code injection. The attack can be initiated remotely. The exploit has been released to the public and may be used for attacks. The vendor was contacted early about this disclosure but did not respond in any way. This vulnerability only affects products that are no longer supported by the maintainer.
-
CVE-2026-115886
MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. From 3.12.0rc0 until 3.17.0, the scorer invoke, register, and GenAI evaluate endpoints, including unauthenticated POST /ajax-api/3.0/mlflow/scorer/invoke, accept serialized third-party scorer data that Scorer.model_validate_json resolves using a module-path allowlist without constraining the selected class or its constructor arguments. When optional trulens and trulens-providers-litellm packages are installed, an attacker can select mlflow.genai.scorers.trulens, choose a model URI outside MLflow's native provider registry so execution uses the LiteLLM fallback, and supply api_base and api_key values that cause the server to send an outbound request to an attacker-chosen or internal host with attacker-controlled headers and body. The vulnerable deserialization occurs before a tracking-store lookup or job submission. This issue is fixed in version 3.17.0.
-
CVE-2026-57845
Hugging Face Datasets before 5.2.0 contains an insecure deserialization vulnerability that allows attackers to execute arbitrary Python code by crafting a malicious pickle file loaded through the deprecated pandas packaged builder. Attackers can supply a pickle-backed dataset that is processed by Pandas._generate_tables via pandas.read_pickle() without validation, so when a victim calls datasets.load_dataset("pandas", data_files=...), the embedded __reduce__ payload executes code with the victim process's privileges.
-
CVE-2026-115749
hydra-ray-launcher 1.1.0.dev1 before 1.2.2 and 1.4.0.dev4 before 1.4.0rc2 contains an arbitrary module import vulnerability via the ray_pkg_version OmegaConf resolver passing input to importlib.import_module(). Attackers controlling configuration values or command-line overrides can supply a ray_pkg_version interpolation to import any installed module, triggering its import-time side effects in the application process.
-
CVE-2025-30047
CGM CLININET is vulnerable to Remote Code Execution caused by deserialization of untrusted data. The server exposes an HTTP-based API, used by the administrative module, that exchanges data using Java object serialization. An attacker within the same local network is able to send requests to this API and transmit a crafted serialized object. When it is deserialized, it allows arbitrary code execution with root privileges on the server. This issue was fixed in version 2026.MS3
-
CVE-2026-114608
Pylint is a static code analyser for Python. From 0.15.2 until 4.0.9, Pylint stores module statistics in predictable files under PYLINT_HOME and loads those files with pickle.load on a later run. An attacker who can write to a shared, group-writable, or otherwise untrusted cache directory can replace a stats cache with crafted serialized data, causing arbitrary code to execute with the privileges of the user who next runs Pylint. The default per-user cache is not normally writable by another user. This issue is fixed in version 4.0.9.
-
CVE-2026-114145
GitPython is a python library used to interact with Git repositories. Prior to 3.2.1, create_submodule() in git/objects/submodule/base.py accepts an attacker-influenced submodule path such as .git/hooks and invokes _clone_repo() before safely rejecting the destination. In 3.1.x, _validate_repo_path() does not exist and the call succeeds silently, while in 3.2.0 and later affected revisions the check in git/index/base.py raises ValueError only after the clone has written to disk and does not remove the planted file. An attacker-controlled submodule can therefore place an executable post-checkout hook that a later Git operation executes. Exploitation requires an application to accept an untrusted submodule path and a clone destination such as .git/hooks that is empty enough for Git to clone into it, which commonly means that default sample hooks are absent. This issue is fixed in version 3.2.1.
-
CVE-2026-92206
CrewAI crewAI Framework Agent Loading Unsafe Reflection Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of CrewAI crewAI. User interaction is required to exploit this vulnerability in that the target must load a malicious agent configuration from the repository. The specific flaw exists within the load_agent_from_repository function. The process does not properly restrict a user-supplied argument before using it to import a module. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-28387.
-
CVE-2026-92202
Luxion KeyShot BIP File Parsing Uncontrolled Search Path Element Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Luxion KeyShot. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the parsing of BIP files. The issue results from loading a library from an unsecured location. An attacker can leverage this vulnerability to execute code in the context of the current process. Was ZDI-CAN-29268.
-
CVE-2026-114096
Dulwich from 0.24.0 before 1.2.8 contains a symlink following vulnerability in porcelain.checkout(paths=[...]) that allows attackers to write arbitrary files by abusing raw os.open without O_NOFOLLOW. Attackers can craft a repository where a worktree symlink points to .git/hooks/post-checkout, so restoring the file from a tag writes a malicious hook for code execution.
-
CVE-2026-91459
Vulnerability in the Oracle Outside In Technology product of Oracle Fusion Middleware (component: Outside In Image Export SDK). The supported version that is affected is 8.5.8. Easily exploitable vulnerability allows unauthenticated attacker with logon to the infrastructure where Oracle Outside In Technology executes to compromise Oracle Outside In Technology. Successful attacks require human interaction from a person other than the attacker. Successful attacks of this vulnerability can result in takeover of Oracle Outside In Technology. CVSS 3.1 Base Score 7.8 (Confidentiality, Integrity and Availability impacts). CVSS Vector: (CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H).
-
CVE-2026-91460
Vulnerability in the Oracle Outside In Technology product of Oracle Fusion Middleware (component: Outside In Image Export SDK). The supported version that is affected is 8.5.8. Easily exploitable vulnerability allows unauthenticated attacker with logon to the infrastructure where Oracle Outside In Technology executes to compromise Oracle Outside In Technology. Successful attacks require human interaction from a person other than the attacker. Successful attacks of this vulnerability can result in takeover of Oracle Outside In Technology. CVSS 3.1 Base Score 7.8 (Confidentiality, Integrity and Availability impacts). CVSS Vector: (CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H).
-
CVE-2026-113730
OGNL is an Object Graph Navigation Library for Java. Prior to 3.4.13, the optional stricter-invocation mode in OgnlRuntime.invokeMethod checks the declaring class of the method that OGNL invokes but does not block java.lang.reflect.Method, java.lang.reflect.Constructor, or java.lang.reflect.Field. An attacker-controlled expression can invoke one of these reflection objects as a proxy for a method on a blocked class, causing the real operation to execute inside JDK reflection without passing through the stricter-invocation check again. Exploitation requires an application to evaluate untrusted OGNL expressions, use the default or an equivalently permissive MemberAccess that permits reflection, and rely on stricter invocation as a sandbox. Applications using a restrictive MemberAccess, including Apache Struts with SecurityMemberAccess, are not affected by this bypass. This issue is fixed in 3.4.13.
-
CVE-2026-112304
docTR through 1.1.0 contains an unsafe deserialization vulnerability in load_pretrained_params() that loads checkpoints with torch.load without weights_only=True. Attackers who control a local or URL-supplied checkpoint can embed a malicious pickle payload to execute Python code with the loading process's privileges on torch releases before 2.6.
-
CVE-2026-112140
A vulnerability was found in llSourcell Doctor-Dignity up to 3f2ffb6c892f13b50a708d05900d3227acb40ded. This impacts the function _get_lib_module of the file python/mlc_chat/chat_module.py. The manipulation of the argument model_lib results in code injection. The attack may be launched remotely. The exploit has been made public and could be used. This product operates on a rolling release basis, ensuring continuous delivery. Consequently, there are no version details for either affected or updated releases. The project was informed of the problem early through an issue report but has not responded yet.
-
CVE-2026-112086
A vulnerability was found in ESPnet up to 202610.post2. This affects the function build_model of the file espnet3/publication/inference_model.py of the component Inference Model Handler. The manipulation of the argument _target_ results in deserialization. It is possible to launch the attack remotely. The exploit has been made public and could be used. The patch is identified as f217b24f06de55e332cab10231469d5890dbbe00. It is best practice to apply a patch to resolve this issue.
-
CVE-2026-112077
A vulnerability was detected in radareorg radare2 6.2.0/6.2.2. This issue affects the function r_core_cmd_str_pipe in the library libr/core/rtr_http.inc.c of the component HTTP Sandbox Policy Handler. Performing a manipulation results in protection mechanism failure. Remote exploitation of the attack is possible. The attack is considered to have high complexity. The exploitability is assessed as difficult. Upgrading to version 6.2.4 is capable of addressing this issue. The patch is named 18fd12efa98a7d1977de7bdd379d8a848777b90f. Upgrading the affected component is recommended. The pre-requisite of the vulnerability is reflected by the CVSS vector. One of the project maintainers explains: "There is a real sandbox configuration issue here, but the report overstates it. The HTTP server defaults to http.sandbox=true, so commands received through /cmd temporarily enable cfg.sandbox. However, cfg.sandbox.grain defaults to all, which means all sandbox capabilities - including exec - remain allowed. Therefore enabling the HTTP sandbox alone does NOT prevent commands from spawning processes. This only matters when the HTTP command endpoint is reachable by someone you do not trust. It is not a remote vulnerability by default: the HTTP server must first be started and exposed to the attacker."
-
CVE-2026-112079
A vulnerability has been found in hankcs HanLP up to 2.1.0-beta.64. The affected element is the function load_word2vec_as_vocab_tensor of the file hanlp/utils/torch_util.py. The manipulation of the argument vectors.vocab leads to deserialization. The attack is possible to be carried out remotely. The exploit has been disclosed to the public and may be used. The project was informed of the problem early through an issue report but has not responded yet.
-
CVE-2026-96620
Claude Agent SDK is a Python SDK for Claude Code. Prior to 0.2.121, the resume option is passed to the Claude CLI as a separate token without validation, so a dash-leading value is parsed as an independent CLI flag. An attacker can inject the --mcp-config flag to define a stdio MCP server, causing arbitrary command execution at CLI startup outside the SDK permission controls. The issue is exploitable when an application passes untrusted input as the session ID to resume. This issue is fixed in version 0.2.121.
-
CVE-2026-111105
Telegram rlottie Numeric Truncation Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Telegram. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the rlottie library. The issue results from the lack of proper validation of user-supplied data, which can result in a write before the start of an allocated buffer. An attacker can leverage this vulnerability to execute code in the context of the current process. Was ZDI-CAN-30207.
-
CVE-2026-110858
GitPython from 3.1.30 before 3.2.0 contains a remote code execution vulnerability in Remote.pull() whose unsafe-option guard inspects only keyword arguments, not the positional refspec. Attackers controlling the refspec can supply --upload-pack=<command> to execute OS commands as the calling process over local-path or SSH transports.
-
CVE-2026-110857
GitPython before 3.2.0 contains a remote code execution vulnerability that allows attackers to run commands because Git.ls_remote forwards URLs without check_unsafe_protocols validation. Attackers can supply ext:: remote-helper URLs to execute commands when git permits the ext transport, while merge_base and IndexFile.move also accept unguarded option injection.
-
CVE-2026-96237
Inclusion of functionality from untrusted control sphere in .NET allows an unauthorized attacker to execute code over a network.
-
CVE-2026-70035
Deserialization of untrusted data in .NET Framework allows an unauthorized attacker to execute code over a network.
-
CVE-2026-108958
A weakness has been identified in TensorLayer up to 2.2.4. This affects the function load_hdf5_graph of the file tensorlayer/files/utils.py of the component HDF5 Loading. Executing a manipulation of the argument model_config can lead to deserialization. The attack can be executed remotely. The exploit has been made available to the public and could be used for attacks. The project was informed of the problem early through an issue report but has not responded yet.
-
CVE-2026-108947
A flaw has been found in yoshitomo-matsubara torchdistill up to 1.1.5. This affects the function load_ckpt of the file torchdistill/common/main_util.py. Executing a manipulation of the argument src_ckpt can lead to deserialization. The attack needs to be launched locally. The exploit has been published and may be used. The project maintainer explains: "This is very similar to an internal report that GH auto-generated for the owner before, which I found tolerable for this project at that time. So I will close this as duplicate."
-
CVE-2026-84318
Nginx UI is a web user interface for the Nginx web server. From 2.0.0-rc.6 until 2.4.1, the MCP tool nginx_config_add writes nginx configuration files to disk and triggers an nginx reload without calling config.ValidateConfigFile. The identical flow in the REST API (POST /api/configs) does call that function and correctly rejects dangerous directives. An authenticated attacker can exploit this gap to write config files containing any blocked directive (perl, js_import, load_module, user root, etc.), bypassing nginx-ui's entire security layer. On a server with the nginx Perl, njs, or Lua module loaded, this is post-authentication Remote Code Execution. This issue is fixed in version 2.4.1.
-
CVE-2026-52903
A deserialization of untrusted data vulnerability was found in ManageIQ. The YamlLoadAliases module overrides YAML.safe_load to silently fall back to YAML.unsafe_load in production when a Psych::DisallowedClass error occurs. An authenticated attacker with dialog import access can exploit this to achieve remote code execution by uploading a crafted YAML payload that triggers the fallback and deserializes arbitrary Ruby objects. Red Hat CloudForms, which was based on ManageIQ, has reached End of Life and is no longer supported. No errata or patches will be issued. This CVE is being assigned for the benefit of the ManageIQ community and downstream consumers.
-
CVE-2026-108820
A weakness has been identified in agno-agi agno up to 3.0.6/3.1.1. This issue affects the function member_agent.run in the library libs/agno/agno/team/_default_tools.py of the component Delegation Feature. Executing a manipulation can lead to execution with unnecessary privileges. The attack can be executed remotely. The exploit has been made available to the public and could be used for attacks. The pull request to fix this issue awaits acceptance.
-
CVE-2026-108816
A vulnerability was found in agno-agi agno up to 3.0.6. Affected by this vulnerability is the function run_python_code of the file python.py of the component PythonTools. The manipulation of the argument code results in code injection. The attack can be launched remotely. The exploit has been made public and could be used. The project was informed of the problem early through an issue report but has not responded yet.
-
CVE-2026-21125
Improper input validation in parsing sheifd in libsimba.heifdec.media.samsung.so prior to SMR Oct-2026 Release 1 allows local attackers to potentially execute arbitrary code.
-
CVE-2026-102554
Allocation of resources without limits or throttling (CWE-770) during Java object deserialization in Google Guava versions 4.0 through 33.7.1 allows an attacker to cause a Denial of Service via OutOfMemoryError. When deserializing CompactHashMap, CompactHashSet, or MapMakerInternalMap instances, Guava eagerly allocates an array based on a caller-specified size parameter without throttling, permitting memory exhaustion from crafted serialization streams.
-
CVE-2026-43598
Improper input validation in the AMD ROCm Communication Collectives Library (RCCL) could allow a compromised peer rank or network-adjacent attacker to dereference an attacker-controlled pointer, potentially resulting in remote code execution.
-
CVE-2026-104809
DigitalCanion has discovered a vulnerability that allows an attacker to cause the system to load an attacker-controlled .so file instead of the expected legitimate module. The loading mechanism relies on a predictable module name without adequately verifying the file’s origin or integrity. A malicious shared object using the expected name can therefore be loaded by a privileged process. The module code then executes within the context and privileges of that process. This results in arbitrary code execution and full compromise of the Mitel Linux virtual machine.
-
CVE-2019-25777
YAML versions before 1.27_001 for Perl allow a loaded perl/glob document to replace any package variable, which can lead to arbitrary code execution. A perl/glob document names a package and a symbol, and supplies the value assigned to it. Nothing restricts the name, so the target can be @INC or YAML's own load options. A perl/glob document that sets $YAML::LoadCode or $YAML::UseCode turns on code loading, which is off by default, for every later Load() in the process. A perl/code document is then passed to a string eval, so an attacker who supplies two documents to separate Load() calls in one process can execute arbitrary Perl code.
-
CVE-2026-64949
Incomplete extension blacklist in the File Manager module allows authenticated upload and execution of arbitrary .phar files. Affects Pandora FMS from 777 onwards.
-
CVE-2026-100841
In MONAI 1.6.0, PersistentDataset (monai/data/dataset.py) explicitly rejects the combination track_meta=True with weights_only=True, forcing users who cache MetaTensors (the default tensor type in MONAI >= 1.0) to run torch.load(hashfile, weights_only=False). Related cache helpers in monai/data/utils.py also call pickle.loads on cached content and derive cache keys with hashlib.md5. As a result, a local user with write access to a shared or world-writable cache_dir (e.g. /tmp/monai_cache, HPC scratch, ~/.cache/monai) can place a malicious pickle file that is deserialized the next time another user's MONAI pipeline reads the cache, resulting in arbitrary code execution in that user's context. All released versions of the monai pip package are affected; no patched version is available as of the advisory.
-
CVE-2026-97230
IO::Socket::SSL::SelfCertificate versions 1.00 for Perl contains malware which executes Python code from an obfuscated URL. The generate_certificate runs a Python script saved as a certificate file. The pyhton script attempts to retrieve code from a hardcoded http URL that is obfuscated with base64 encoding and run the response body directly. The impact is that arbitrary code can be invoked as the user, without a dropped script being saved on the affected host. The releases have no test scripts nor build hooks. The intention may have been to trigger the payload after installation. The dropper script is in lib/Crypt/SelfCertificate/sample/cert.pem. This is similar to CVE-2026-95831 for the module Crypt::SelfCertificate. The SHA-256 digests of the files are ba24ee8ec3b7f47f65bed62e16fb413ace50653cf44bd8ea90914390922831e0 IO-Socket-SSL-SelfCertificate-1.00.tar.gz 821d38830e5eb8607738421c25ac25f59fff02a6ab67daa32fbd020429454dac IO-Socket-SSL-SelfCertificate-1.00/lib/IO/Socket/SSL/SelfCertificate/sample/cert.pem d483cb7b23b7271cb11cf242bff4a2e1c02df0b9525eb0429abeea8961c399d5 IO-Socket-SSL-SelfCertificate-1.00-upload.tar.gz
-
CVE-2026-84502
A flaw was found in Red Hat Ansible Automation Platform's automation- controller. The Project scm_url field is not validated against values that begin with a dash and is stored and passed verbatim to the git SCM module. Because the module runs git ls-remote with the URL as a positional argument and without a "--" separator, a git project URL such as "--upload-pack=<command>:x" is interpreted by git as the --upload-pack option and executed via a shell. A user with permission to create or modify a project in a single organization can thereby execute arbitrary commands on the control-plane task pod, with output reflected through the project update stdout endpoint, leading to cross-tenant compromise and in-cluster lateral movement
-
CVE-2026-91813
A vulnerability in Foxit PDF Editor/Reader’s update mechanism allows an update package to be replaced between download and high-privilege extraction due to insufficient file locking and integrity validation. This could enable local attackers to execute arbitrary code with elevated privileges.
-
CVE-2026-86805
A time-of-check to time-of-use (TOCTOU) race condition in the dynamic loader (ld.so) of the GNU C Library (glibc) versions 2.14 through 2.44 allows a local attacker to escalate privileges. When expanding $ORIGIN in DT_RPATH for setuid/setgid (AT_SECURE) programs, glibc validates the lexically normalized search path against the trusted directories but then opens the raw, un-normalized path. On systems where the Linux fs.protected_hardlinks sysctl is disabled, a local attacker who hard-links such a program into an attacker-controlled directory and wins a race to replace an intermediate path component with a symbolic link can direct the loader outside the trusted directory, causing it to load an attacker-controlled shared object and execute arbitrary code with the elevated privileges of the program. Exploitation requires an installed setuid or setgid binary whose DT_RPATH uses $ORIGIN followed by ".." traversal that normalizes into a trusted directory, and the ability to hard-link that binary and win the race by swapping a path component for a symbolic link. Major Linux-based OS distributions ship with fs.protected_hardlinks enabled by default and mitigate the vulnerability.
-
CVE-2026-94383
The MISP blocklist workflow module accepted a user-supplied blocklist filename parameter without validating the file extension. The only sanitization applied was basename() to strip path components and a check for empty or dot values. A site administrator could specify a filename with an arbitrary extension that would be placed in the MISP export directory. If the underlying web server is configured to interpret and execute scripts from that directory, the resulting file could be invoked, leading to arbitrary code execution in the context of the web server process. The vulnerability requires the attacker to hold site-administrator privileges within MISP, as the blocklist workflow module is restricted to that role. No additional user interaction is required beyond triggering the workflow action with a crafted filename parameter. The impact is full compromise of the MISP server's confidentiality, integrity, and availability, as arbitrary script execution grants the attacker the same privileges as the web server user. Version affected: <2.5.47
-
CVE-2026-93606
vm2 (npm) versions 3.12.0 and earlier contain a sandbox escape in `VM` and `NodeVM`. When an embedder exposes a host API that returns a host-realm Promise, the bridge's rejection sanitizer (hostPromiseSanitizeReject / makeSanitizedPromiseCallback / normalizeHostPromiseCallbacks in lib/bridge.js) only wraps `then`/`catch` rejection slots that hold a function, and the sandbox-side `Symbol.species`/`.then` neutralization is installed only on the sandbox intrinsic `Promise.prototype`, so it never applies to a host Promise. Code running inside the sandbox can overwrite `p.constructor[Symbol.species]` on the host Promise and then call `p.then()` with no `onRejected` handler; V8 substitutes its internal Thrower, which re-throws the raw host rejection value into a resolve/reject closure captured by the attacker. This delivers an unsanitized, fully functional bridge proxy of the host object to sandboxed code, bypassing handleException and hostPromiseSanitizeReject. If the rejection value is host-pivotable (for example a host `process` object), this results in arbitrary code execution on the host. Fixed in 3.12.1.
-
CVE-2026-54916
NetBox Device Type Library is a collection of community-sourced device type definitions for import into NetBox. The absence of tests/init.py and the lack of --import-mode=importlib cause pytest prepend import mode to place the tests directory at the front of sys.path during collection. An unauthenticated contributor can add a module such as tests/git.py that shadows GitPython when tests/definitions_test.py executes from git import Git, Repo, or add tests/conftest.py for automatic collection-time execution. Python imports and runs the pull-request module before any test function, allowing arbitrary code execution on the GitHub Actions runner, test-result tampering, and access to tokens or network resources exposed to the workflow. This module-shadowing path is independent of the earlier pickle deserialization flaw and the separately tracked NETBOX_DT_LIBRARY_URL issue. This vulnerability is fixed by commit b0d9a3dadd0a0a9d3c93b0b2777559fd4bad1037.
-
CVE-2026-48977
OpenSlide is a C library for reading whole slide image files. From 3.4.1 until 4.0.1, OpenSlide's parse_level0_xml() processing in src/openslide-vendor-ventana.c accepts nonpositive row or column tile counts from a crafted Ventana BIF file. The invalid counts produce attacker-controlled relative memory offsets and allow arbitrary values to be written at those offsets, affecting all supported platforms and configurations and resulting in a crash or potential arbitrary code execution. This issue is fixed in version 4.0.1.
-
CVE-2026-54752
NetBox Device Type Library is a collection of community-sourced device type definitions for import into NetBox. The validation test harness can deserialize pull-request-controlled tracked pickle cache files through pickle.load in the read_pickle_data function in tests/pickle_operations.py. An unauthenticated contributor can change USE_LOCAL_KNOWN_SLUGS in tests/test_configuration.py and supply a crafted tests/known-modules.pickle or tests/known-racks.pickle file that tests/definitions_test.py loads when pytest runs. Deserialization invokes attacker-controlled object reduction behavior, allowing arbitrary code execution in the GitHub Actions runner or in a maintainer process that runs the tests, with the confidentiality, integrity, and availability of reachable resources at risk. This vulnerability is fixed with commit 1c6f7e2b93589b965318c6e67ac3504831f0e71e.
-
CVE-2026-52727
lxc-ci contains continuous integration and image-build scripts for LXC. Prior to the 2026-05-28 Arch Linux image publication, images built from images/archlinux.yaml retain the same pacman local-signing private key in /etc/pacman.d/gnupg and redistribute it to every container or virtual machine created from that image. An attacker who controls an HTTP package mirror or can intercept mirror traffic can use the shared pacman signing private key to sign modified packages that affected clients accept as trusted. Installing those packages permits arbitrary code execution as root on the client system. This issue is fixed in Arch Linux images published on or after 2026-05-28.
-
CVE-2026-92946
vm2 before 3.11.7 contains a remote code execution vulnerability when require.external is enabled without an explicit require.root that excludes node_modules. Sandboxed code can require vm2's own package, instantiate an unrestricted NodeVM instance, and execute arbitrary host OS commands via child_process.
-
CVE-2026-92935
vm2 is a sandbox for running untrusted Node.js code. In versions >= 3.11.4 and <= 3.11.6, the NodeVM constructor computes `hasRealRequireConfig` with `typeof requireOpts === 'object' && requireOpts !== null`, so an array-shaped `require` value (for example `require: []`) satisfies the guard that is meant to reject nesting without an explicit require configuration. `makeResolverFromLegacyOptions()` then destructures the array into undefined option fields and returns a resolver containing only `NESTING_OVERRIDE.vm2`. As a result, an attacker who can supply JavaScript executed by a NodeVM configured with truthy `nesting` and an array-shaped `require` (e.g. `new NodeVM({nesting: true, require: []})`) can require the host `vm2` module, create an inner NodeVM with an attacker-chosen builtin allowlist (such as `child_process`), and execute arbitrary commands with the privileges of the host Node.js process, escaping the sandbox. Outer builtin restrictions do not constrain the attacker-created inner NodeVM. This issue is fixed in vm2 3.11.7.
-
CVE-2026-92838
A DLL hijacking vulnerability exists in the GeoVision GV-Remote E-Map desktop application. The application loads one or more dynamic-link libraries (DLLs) from an unsafe search path, allowing a local attacker to place a malicious DLL in a location searched before the legitimate library location. If successfully exploited, an attacker with local write access to the affected directory could achieve arbitrary code execution in the security context of the GV-Remote E-Map process.
The newest published CVE records from CVE.org and NVD, refreshed every few minutes, with severity, exploit odds and fix status.