Vulnerabilities

Tesseract OCR versions 5.5.3 and earlier contain a heap out-of-bounds write (CWE-787) in the LSTM recognition path: while earlier hardening added bounds checks to NetworkIO::CopyTimeStepGeneral and NetworkIO::Randomize, the NetworkIO::WriteTimeStepPart and NetworkIO::AddTimeStepPart functions in src/lstm/networkio.cpp remained unchecked. The flaw is triggered when the default LSTM engine processes a specially crafted .traineddata model file: in LSTM::Forward (src/lstm/lstm.cpp) the source_ buffer is sized from the independently deserialized na_ field, but WriteTimeStepPart writes a count driven by ns_ from the CI gate WeightMatrix dim1() value, so a crafted NT_LSTM layer with ns_ much larger than na_ overruns the heap buffer during the first recognition step. An attacker who can get such a model file processed gains heap corruption, causing a crash or potentially controlled memory corruption; CVSS 4.0 rates this 8.6 High with a local attack vector and no privileges or user interaction required. Anyone running Tesseract 5.5.3 or earlier is affected in principle, with practical risk concentrated in applications and pipelines that load third-party or otherwise untrusted .traineddata language models rather than only official trusted ones. No public proof-of-concept or in-the-wild exploitation is known, the flaw is not in CISA KEV, and no fixed release was available at the time of review.

Do: No fixed Tesseract release is available yet, so monitor the project for a patched version and upgrade as soon as one ships. Until then, avoid loading .traineddata files from untrusted or third-party sources on affected builds, verify the provenance and integrity of any externally sourced language models, and consider running OCR of external models in sandboxed, low-privilege workers to limit impact of a crash or heap corruption.

MISP versions 2.5.45 and earlier accept attacker-controlled email input on the unauthenticated forgot-password reset endpoint without a length bound or format validation, persisting the supplied value twice per request (once in an audit log entry and once in a queued password-reset job) with no rate limiting. An unauthenticated attacker can repeatedly submit oversized or arbitrary values to this and similar pre-auth paths (registrations, failed REST authentication attempts), driving unbounded growth of persistent storage and queued work. The result is high availability impact (CVSS 4.0 8.7, VA:High): resource exhaustion can degrade or disrupt the instance, while confidentiality and integrity are unaffected. Any MISP deployment running 2.5.45 or earlier is affected, particularly instances where the pre-auth reset flow is reachable by untrusted users. No public PoC, no CISA KEV listing, and an EPSS of about 0.3% indicate no known exploitation at this time.

Do: Upgrade MISP to a release newer than 2.5.45, which adds the 1024-byte email length cap, email-format validation before persistent work, a per-source pre-auth request budget with HTTP 429 responses, and a 15-minute cooldown for API-access request emails. Until patched, rate-limit or restrict unauthenticated access to the forgot-password and other pre-auth endpoints at a reverse proxy or WAF, and monitor audit-log and job-queue growth for signs of storage exhaustion. Also verify the API-access request endpoint is POST-only with CSRF protection in the deployed version.

CVE-2026-86218 is a static code injection flaw (CWE-96) in N-able's N-central on-premises remote monitoring and management (RMM) platform, carrying a maximum CVSS 4.0 score of 10.0. An unauthenticated, remote attacker triggers it by sending crafted network input to the N-central server that is improperly neutralized and persisted into application-managed code, which the server then executes — no privileges (PR:N) or user interaction (UI:N) are required. Successful exploitation yields full server compromise with high impact on confidentiality, integrity, and availability, and because N-central acts as the management hub for downstream customer endpoints, compromise can expose the entire managed estate. Any organization running an affected N-central release (before 2026.3.1.14) — primarily MSPs and corporate IT departments using N-able RMM — is affected. The flaw is confirmed exploited in the wild: N-able patched it as a zero-day, CISA added it to the KEV catalog on 2026-09-08, and it is the fourth N-central hotfix in five weeks, though no public PoC is known and ransomware use is unknown.

Do: Upgrade N-central to 2026.3.1.14 or later (or apply N-able's hotfix) immediately, as the flaw is in CISA's KEV catalog and BOD 26-04 timelines apply to federal stakeholders. Until patched, remove direct internet exposure of the N-central server (restrict to VPN/management networks via firewall allowlists) since no authentication is needed for exploitation. Because in-the-wild exploitation is confirmed, review internet-facing N-central servers for indicators of compromise such as unexpected processes, unusual child processes of the web service, and new or suspicious accounts.

MISP's custom LdapAuthenticate and LinOTPAuthenticate components, which replace CakePHP's FormAuthenticate, failed to replicate its credential validation, allowing empty or non-string values to reach the underlying authentication mechanisms. A remote attacker who knows a valid directory user's email address can submit an empty password, which is passed to ldap_bind(); if the LDAP server accepts unauthenticated binds, the bind succeeds and MISP treats the attacker as that directory user without verifying any password. The flaw also affects the LinOTP path and local fallback: with mixed authentication enabled, an empty password can be checked against a stored MISP password hash, and LDAP-provisioned accounts may have been created with an empty local password that remains usable once the account is no longer resolved through LDAP. Successful exploitation lets an unauthenticated attacker impersonate an existing MISP user — potentially an administrator — gaining access to sensitive threat-intelligence data and the ability to modify or delete it and alter configuration. Only MISP deployments with the LDAP or LinOTP authentication components enabled are affected; no public proof-of-concept or in-the-wild exploitation is known (EPSS 0.5%, not in CISA KEV).

Do: Upgrade MISP to the patched release addressing CVE-2026-85216; the fix enforces valid string credentials, rejects empty passwords where not permitted, and assigns randomly generated local passwords to LDAP-provisioned accounts. As interim mitigations, configure the LDAP directory to reject unauthenticated (empty-password) binds, disable mixed authentication with LinOTP if not needed, and reset the local passwords of LDAP-provisioned MISP accounts — especially privileged ones — in case they hold empty-password hashes. Review authentication logs for successful logins against known directory-user email addresses using empty passwords.

CVE-2026-19586 is a pre-authentication OS command injection flaw (CWE-78) in TP-Link Omada business gateways when they are configured to operate as an OpenVPN Server, caused by insufficient validation of client-supplied data during OpenVPN connection establishment. An unauthenticated remote attacker who can reach the VPN service can send specially crafted input during a connection attempt to influence backend command execution before authentication completes, gaining arbitrary command execution and potentially full compromise of the gateway, which typically sits at the network edge controlling routing and VPN for the whole site. Affected organizations are those running any of the listed Omada gateway models with the OpenVPN Server feature enabled and reachable by the attacker. TP-Link rates the issue critical (CVSS 4.0 base score 9.3). A public technical write-up/PoC exists and EPSS assigns a 5.7% probability of exploitation within 30 days (93rd percentile), but exploitation has not yet been confirmed in the wild and the flaw is not in CISA KEV.

Do: Update all listed Omada gateway models to the latest firmware per TP-Link's security advisory (fixed version numbers are not specified in the available data, so check the advisory for affected/fixed ranges). Until patched, disable OpenVPN Server or restrict access to the VPN service (e.g., firewall/ACL rules limiting UDP 1194 to trusted sources). Inventory your estate for these gateway models, confirm whether OpenVPN Server is enabled and reachable, and review VPN logs for anomalous connection attempts.

An unauthenticated denial-of-service vulnerability was identified in TP-Link TL-MR100 v3.2, TL-MR150 v3.2, TL-MR6400 v8.0 and Archer MR600 v2, due to improper handling of exceptional request conditions that may lead to a NULL pointer dereference. A remote attacker on an adjacent network can send a specially crated HTTP request to trigger a crash of the HTTP service process. Successful exploitation may cause the HTTP service to crash, making the web management interface and HTTP-dependent functionality temporarily unavailable.

An OS command injection vulnerability exists in the web management interface of Archer C20 v6 firmware when processing certain WAN-related configuration operations. An authenticated administrator may exploit insufficient input validation to execute arbitrary system commands, potentially resulting in full device compromise. Successful exploitation may allow arbitrary command execution with elevated privileges, compromising the confidentiality, integrity, and availability of the affected device and network traffic passing through it.

Tapo C100/C101 V5 contains a null pointer dereference vulnerability in the RTSP service. An attacker on the local network can send specially crafted requests that cause the service to dereference an invalid pointer, resulting in a service crash and device reboot. Successful exploitation can disrupt live video streaming functionality and cause a temporary denial-of-service condition.

CVE-2026-15316 is an improper input validation flaw (CWE-20) in the configuration service of the TP-Link Tapo C200 camera (v5) that processes encrypted credential data. An attacker on an adjacent network can send oversized ciphertext values with no authentication required; insufficient validation causes exception-handling failures that crash or restart the device. The impact is a denial-of-service condition that temporarily disrupts HTTPS management and monitoring until the service recovers, with no confidentiality or integrity impact per the CVSS vector. Anyone running a Tapo C200 v5 camera is affected, though exploitation requires the attacker to reach the device's network (adjacent-network vector). No public proof-of-concept exists, the issue is not in CISA KEV, and EPSS estimates only a 0.2% chance of exploitation within 30 days.

Do: Check your Tapo C200 firmware version in the Tapo app and apply TP-Link's firmware update for v5 when released. In the meantime, keep the camera's management interface off guest/WAN-exposed network segments and restrict it to trusted LANs, since the attack is unauthenticated and adjacent-network. Note that related reporting on TP-Link camera flaws (including eavesdropping issues) makes prompt patching of Tapo devices generally advisable.

TP-Link Tapo C120 (v1) and Tapo C200 (v5) cameras contain an improper authentication vulnerability (CWE-287) in the login authentication verification module, rated high severity at CVSS 4.0 8.7. An attacker already present on the same local network as the camera (adjacent-network attack vector) can exploit weak validation of challenge parameters during the login handshake to bypass normal authentication and obtain administrative session tokens. With these tokens, the attacker can perform privileged management actions, gain unauthorized administrative access to the camera, and temporarily disrupt device services, causing a denial-of-service condition. All consumer and small-business deployments of these specific camera hardware/firmware versions are affected. A public proof-of-concept exists on GitHub, but EPSS is low (0.3% in 30 days) and the flaw is not in the CISA KEV catalog, so exploitation in the wild is not currently observed.

Do: Check the hardware/firmware version of any Tapo C120 or C200 units in your environment (Tapo app > device settings > firmware) and apply the latest firmware TP-Link publishes for these models as soon as a fix is available, monitoring TP-Link's security advisory page. Because exploitation requires local network access, place cameras on a segregated IoT/guest VLAN that cannot reach trusted internal segments, and verify camera accounts, bindings, and recorded footage for signs of unauthorized administrative access.

The web interface of the affected device relies on the HTTP referrer header as part of request validation. Requests containing empty Referer value, or omitting the Referer header entirely, may be accepted and processed due to insufficient validation logic. Successful exploitation may allow an adjacent attacker with access to the web management interface to obtain device configuration details and other sensitive information.

Ash Framework (the ash Elixir package) before 3.31.3 fails to safely deserialize the client-supplied keyset pagination cursor (page[:after] or page[:before]) consumed by read actions. The :safe option passed to non_executable_binary_to_term/2 blocks new atoms, functions, and ports, but it does not block structs built from atoms already loaded in a running Ash application, so an attacker can forge a cursor containing an %Ash.Query.Call{} expression that is spliced into the keyset filter without the normal public/private expression gating. On the AshPostgres data layer the injected call is inlined into the SQL query, yielding SQL injection; on the ETS and Simple data layers it is evaluated in-process, yielding arbitrary code execution. Any application running ash 1.17.0 through 3.31.2 that exposes keyset-paginated read actions to untrusted clients is affected, with impact depending on the data layer in use. No confirmed in-the-wild exploitation is reported; a public advisory with a proof of concept (GHSA-3gq3-9xm3-c8v3) is available and EPSS estimates roughly a 0.8% chance of exploitation within 30 days.

Do: Upgrade to ash 3.31.3 or later, which is the first fixed release. As an interim mitigation, restrict keyset pagination parameters (page[:after]/page[:before]) on read actions to trusted callers and validate cursor input before it reaches the keyset decoder. Prioritize patching where AshPostgres backs unauthenticated read actions (SQL injection risk) or where ETS/Simple data layers are used (code execution risk), and review GHSA-3gq3-9xm3-c8v3 for the published proof of concept.

ScadaLTS 2.7.8.1 exposes a server-side method that lacks authorization checks, allowing any authenticated user (including one holding only low-privilege, read-only permissions) to execute arbitrary operating system commands on the host. Successful exploitation results in code execution in the context of the ScadaLTS server process (root), leading to full compromise of the underlying system.

Uncontrolled Resource Consumption vulnerability in ash-project ash allows an attacker to exhaust the memory of the node via a crafted keyset pagination cursor. Read actions with keyset pagination deserialize the client-supplied page[:after] or page[:before] cursor in decode_values/2 in lib/ash/page/keyset.ex, which base64-decodes the value and passes it to :erlang.binary_to_term/2 without bounding its size. The Erlang external term format supports zlib-compressed payloads, which the decoder inflates transparently, so a cursor of a few kilobytes can allocate tens of megabytes of heap in a single call. Ash itself only ever encodes cursors uncompressed, so the decoder accepts a term shape its encoder never produces. Concurrent requests aggregate these allocations and can terminate the node. This issue affects ash: from 1.17.0 before 3.31.1.

CVE-2026-66838 is a SQL injection (CWE-89) in the Elixir PostgreSQL driver Postgrex, affecting versions from 0.19.3 up to but not including 0.22.4. The flaw arises because Postgrex.stream/4 embeds the caller-supplied :comment option into the statement text of the Parse message by raw concatenation, without invoking the comment_not_present!/1 validation applied at every other execution point, so an attacker who can influence that value can close the comment with */ and append their own SQL clauses, which execute under the database connection's role. Applications that pass attacker-controlled data as :comment — including through Ecto.Repo.stream/2, which exposes the same option — are affected; because Parse accepts only a single command, the injection is confined to the streamed statement and further statements cannot be chained. CVSS 4.0 rates this 5.9 (medium) with a precondition that the attacker has a way to influence the comment value, and no exploitation in the wild is currently known (EPSS 0.3%, not in CISA KEV), though a public advisory with a proof-of-concept exists (GHSA-3gww-3f36-2388).

Do: Upgrade postgrex to 0.22.4 or later via your dependency manager, and review the public advisory GHSA-3gww-3f36-2388. If you cannot upgrade immediately, audit whether your code calls Postgrex.stream/4 or Ecto.Repo.stream/2 with a :comment option and, if so, ensure that value never originates from untrusted input or sanitize it to reject the */ sequence. Because exploitation requires application code to forward attacker-influenced data into :comment, checking that single call pattern is the fastest way to determine actual exposure.

Tapo P110 v1 smart Wi-Fi Plug contains an improper boundary validation vulnerability in the handling of authenticated HTTP request bodies due to insufficient input validation before memory copy operations. This may lead to buffer overflow condition, causing the web service process to crash. Successful exploitation may cause the web service process to stop responding or restart, resulting in a denial-of-service condition.

A certification validation weakness exists in communication between affected Omada devices and cloud controllers. Certificate identity verification does not adequately validate that a presented certificate corresponds to the expected cloud controller hostname, which may allow certificate validation protections to be bypassed under specific conditions. Successful exploitation may allow interception or modification of communication between affected devices and cloud controllers.

Affected Omada devices rely on embedded certificates that are shared across deployments to establish trust between controllers and managed devices. An attacker who obtains the embedded certificates may be able to impersonate trusted controllers or devices and intercept affected communications.

OpenEMR through 8.2.0 contains a command-execution flaw in its document category tree component (library/classes/Tree.class.php), where values from the categories database table are passed into an unsanitized eval() call (CWE-95). An attacker who has obtained administrator-level access and can execute arbitrary SQL alters the id column of the categories table to VARCHAR and inserts a PHP payload, which is then executed whenever any page instantiates CategoryTree, including unauthenticated and low-privilege pages. Successful exploitation yields arbitrary operating-system command execution as the web server user, providing a persistent foothold until the database tampering is cleaned up. Any OpenEMR deployment running version 8.2.0 or earlier is affected, including self-hosted clinic and health-center installations. Exploitation has not been confirmed in the wild, but two public proof-of-concept write-ups exist, the flaw is rated 9.4 (Critical), EPSS currently estimates a 3.7% probability of exploitation within 30 days (89th percentile), and the issue is not in CISA's KEV catalog.

Do: Upgrade OpenEMR to a release newer than 8.2.0 as soon as the project publishes a patched version, prioritizing internet-exposed instances. Until patched, audit the categories table for tampering (an id column changed to VARCHAR or unexpected stored values), remove arbitrary-SQL capability from untrusted administrator accounts, and restrict the web server user's OS command permissions. Given the public PoCs and rising EPSS, monitor logs for anomalous requests to pages that instantiate CategoryTree and for suspicious commands run as the web server user.