Unauthenticated information disclosure in Evolution API through 2.3.7: an incorrect array comparison in the metricsIPWhitelist middleware always evaluates to false, so the IP allowlist check never enforces and any network attacker can reach the Prometheus /metrics endpoint. The exposed metrics leak server version, database client name, configured server URL and WhatsApp instance details, which is useful reconnaissance material for follow-on attacks against self-hosted WhatsApp gateways. Publicly available exploit code exists (VulnCheck-reported, with a linked GitHub write-up); the issue is not listed in CISA KEV and no EPSS score was provided in the input.
SQL injection in the WP Directory Kit WordPress plugin (versions through 1.5.7) lets authenticated users with page-builder access (Editor role and above) inject arbitrary SQL through widget settings that are stored without sanitization or escaping and then concatenated into a database query. The injected statement executes when the affected page is rendered, allowing the attacker to read arbitrary data from the WordPress database. Publicly available exploit code exists (WPScan advisory) and the vendor CPE is still listed as 'unknown', but there is no CISA KEV entry and EPSS is only 0.16%, indicating targeted rather than opportunistic mass exploitation.
OS command injection in SabyasachiRana WebMap's nmap_newscan() function permits remote attackers to execute arbitrary shell commands by injecting metacharacters through the unsanitized target and params arguments in the New Nmap Scan Handler (functions_nmap.py). All repository commits up to 8b95fe4dc301a3c09ddf145b895de0bf9f8d2a25 are affected, and the flaw is exploitable without authentication per the CVSS PR:N vector. A public proof-of-concept exploit is available and a silent upstream patch has been committed without a formal security advisory.
Unauthenticated access to the `/blacklist` endpoint of Maltrail (versions up to 3.0.1) allows any remote caller to retrieve source IP addresses of events that matched the operator's BLACKLIST rules, exposing internal network hosts flagged by the threat detection system. The oversight - a missing session check on one endpoint while all sibling endpoints (`/events`, `/counts`, `/hunt`, `/meta`, `/geo`) correctly required authentication - was confirmed by the vendor and patched in version 3.1 the same day it was reported. A public proof-of-concept exploit is available, making exploitation trivially scriptable against exposed instances.
Unauthenticated information disclosure in D-Link DIR-882 routers (firmware up to 20260814) allows remote attackers to retrieve sensitive data by sending crafted requests to the /HNAP1/dllog.cgi CGI binary endpoint. The vulnerability resides in the main() function of the dllog.cgi component, which fails to properly restrict output before returning it to the caller. A public proof-of-concept exists, per the GitHub reference, which also links this endpoint to quick VPN settings disclosure - suggesting credentials or configuration material may be exposed. No public KEV listing and no vendor patch have been identified at time of analysis.
Use-after-free in GPAC's LASeR scene graph command processing subsystem allows remote attackers to corrupt heap memory by supplying a crafted media file, potentially leading to process crash or code execution. Affected are GPAC builds up to commit f1219cde; the flaw stems from direct node-pointer assignment without proper reference counting in the LASeR decoder and scene graph command application paths. A public proof-of-concept exploit is confirmed available; vendor-released patch exists at commit e34f4ba and release abi-16.24.
Use-after-free in zstd-jni's Dictionary Sharing component allows remote attackers to corrupt native heap memory through the ZstdCompressCtx.loadDict() and ZstdDecompressCtx.loadDict() APIs in versions up to 1.5.7-13. The defect is a logic inversion in shared-lock accounting: the code releases the reference count on the incoming dictionary parameter rather than on the previously held dictionary reference, causing the old native buffer to remain referenced after it can be freed by another thread. A public proof-of-concept exists via GitHub issue #404, and vendor released a confirmed fix in version 1.5.7-14.
Authorization bypass in subhajitkhan's online-clinic-management-system lets a remote, unauthenticated attacker take over an administrator session by manipulating the adminmail argument processed by session_start() in adminappview.php, with no password or token required. Publicly available exploit code exists and the maintainer has not responded to the issue report filed against the project, so no fixed release is available. With CVSS 4.0 base metrics of AV:N/AC:L/PR:N/UI:N and E:P, impact is scored low across confidentiality, integrity and availability, but the resulting admin session can expose patient and appointment records.
Unauthenticated SQL injection in SourceCodester Online Faculty Clearance System 1.0 allows remote attackers to inject arbitrary SQL through the ID argument of /delete_requirement.php, giving them read and write access to the application's clearance records. Publicly available exploit code exists (GitHub OnlyCMJB/cve issue 2), and the flaw is tracked by VulDB and ENISA EUVD, but CISA KEV does not list it as actively exploited. Because this is a demo-grade PHP application distributed as downloadable source with no vendor patch channel identified, exposure is concentrated among institutions that deployed it as-is on an internet-facing web server.
Unauthenticated SQL injection in the login handler (/College/login.php) of SourceCodester College Notes Gallery Management System 1.0 allows remote attackers to inject arbitrary SQL through the 'User' parameter. Publicly available exploit code exists (disclosed via a GitHub issue referenced by VulDB), and successful abuse can bypass authentication and expose or corrupt the backend database. The CVSS 4.0 base score is 5.5 with the E:P threat metric, reflecting limited but real confidentiality, integrity, and availability impact.
Unauthenticated remote SQL injection in SourceCodester Online Faculty Clearance System 1.0 allows an attacker to inject arbitrary SQL through the ID argument of /delete_faculty1.php, exposing or destroying records in the underlying database. The CVSS 4.0 vector rates network-reachable, low-complexity, no-privilege, no-interaction exploitation with low confidentiality, integrity and availability impact (base 5.5), and publicly available exploit code exists, so unpatched internet-facing instances are immediately at risk. No CISA KEV entry or EPSS score was supplied, so there is no confirmed evidence of active exploitation in the wild at time of analysis.
SQL injection in zyx0814 FilePress up to version 3.0.1 allows remote unauthenticated attackers to manipulate database queries via the orderby and order parameters in dzz/publish/search.php. The Publish Module's search endpoint accepts unsanitized user-controlled sorting parameters directly into a SQL ORDER BY clause, enabling blind or time-based data extraction. A publicly available proof-of-concept exploit exists; the vendor has not responded to the coordinated disclosure report as of the time of analysis.
SQL injection in PHPGurukul Daily Expense Tracker System 1.1 exposes the unauthenticated forgot-password endpoint `/dets/forgot-password.php`, where the `email` and `contactno` parameters are passed unsanitized to backend SQL queries. Any remote, unauthenticated attacker can exploit this to read or manipulate the underlying database. A public proof-of-concept exploit has been disclosed on GitHub, lowering the skill threshold for exploitation; no KEV listing exists at time of analysis.
SQL injection in PHPGurukul Daily Expense Tracker System 1.1 allows unauthenticated remote attackers to manipulate the email parameter in the pre-authentication login form at /dets/index.php, enabling authentication bypass or database content extraction. A public exploit is available on GitHub, removing any meaningful skill barrier to exploitation. No vendor-released patch has been identified; the CVSS 4.0 score of 5.5 with all impacts rated Low reflects the application's narrow data scope rather than a general downgrade of the SQL injection risk class.
Denial of service in Open5GS 2.7.7 and earlier lets a remote attacker crash the SGW-U/UPF process by sending a crafted PFCP Session Establishment or Modification Request containing a malformed Outer Header Creation IE. The handler fails to validate the Outer Header Creation description bits and payload length, so a failed GTP-U peer connection leaves a node with an unset address family in the peer list; a subsequent request referencing the same address reaches ogs_pfcp_far_f_teid_hash_set and aborts the process. Publicly available exploit code exists (GitHub issue #4699) and an upstream fix commit has been published, but the flaw is not listed in CISA KEV.
Unauthenticated SQL injection in WuzhiCMS 4.0 and 4.1.0 allows remote attackers to tamper with backend database queries through the article::getDataOfJson action, because the title and master_table parameters are passed into SQL statements without sanitization. Publicly available exploit code exists (referenced via the vendor's own GitHub issue tracker), and the upstream project has not responded or issued a fix at the time of analysis. The CVSS v4.0 score of 6.9 with an E:P threat metric reflects confirmed proof-of-concept availability rather than an abstract theoretical risk.
GitLab CE/EE instances that enforce SAML single sign-on can be bypassed by an authenticated user who signs in through another authentication path, because the sign-in enforcement check is missing (CWE-306). Affected releases are all versions from 18.6 up to 19.1.7, 19.2.x through 19.2.5, and 19.3.0/19.3.1; GitLab has published fixes in 19.1.8, 19.2.6 and 19.3.2. Publicly available exploit code exists (a HackerOne report is linked), but the issue is not listed in CISA KEV and no active exploitation has been confirmed; CVSS is a moderate 5.4 and impact is limited to partial confidentiality and integrity loss, not remote code execution or full account takeover of the platform.
WP Directory Kit, a WordPress directory/listing plugin, exposes the content of draft and unapproved listings to anyone who can reach its public AJAX actions. Versions through 1.5.7 omit status and ownership checks before returning listing content, so unauthenticated attackers can enumerate other users' unpublished submissions (CWE-200, information disclosure). Publicly available exploit code exists per the WPScan reference, though EPSS is only 0.14% (4th percentile) and there is no CISA KEV listing, so this is a moderate, opportunistic-targeting issue rather than a mass-exploitation event.
Improper authorization checks on GitLab CE/EE internal data emission endpoints let an authenticated user retrieve sensitive credentials and tokens that should only have been returned along the expected proxy path. All releases from 10.1.0 before 19.1.8, the 19.2 branch before 19.2.6, and the 19.3 branch before 19.3.2 are affected according to the vendor, and publicly available exploit code exists (HackerOne report 3907835). Impact is confidentiality-only (CVSS 5.3, C:L/I:N/A:N), with no CISA KEV listing and no EPSS score supplied in the source data.
Jpom through 2.11.12 exposes a broken object-level authorization flaw on the /build/branch-list endpoint, letting any authenticated user supply a repositoryId belonging to a different workspace and have the server resolve it without verifying ownership. This enables cross-workspace repository enumeration, disclosure of repository type, and execution of git ls-remote calls that reuse another workspace's stored Git credentials, leaking private repository metadata. Exploitation is low-complexity over the network with only a low-privileged account required, and publicly available exploit code exists, though the CVSS 4.0 base score is only 5.3 because the impact is limited to confidentiality.
CSRF protection bypass in MISP ≤2.5.45 allows a network-adjacent attacker to forge authenticated state-changing requests against any MISP action that reads parameters from the URL. The bypass exploits CakePHP's method-override processing: submitting `_method=GET` (or an equivalent non-write verb via `X-HTTP-Method-Override`) causes CakePHP to empty the parsed request body before MISP's `BetterSecurityComponent` evaluates whether to run `_validatePost()` and `_validateCsrf()`. Because `hasData` is false after body-clearing, both protections are silently skipped entirely. No public exploit has been identified at time of analysis, but the mechanism is straightforward to weaponize by anyone who can lure an authenticated MISP user to a crafted page.
Denial of service through catastrophic regular-expression backtracking affects the A2UI web_core and React renderer packages through version 0.10.6, where the Basic Catalog 'regex' function and TextField validation compile and execute attacker-supplied patterns without any safety check. An application that renders attacker-influenced A2UI schemas or expressions can have its JavaScript event loop pinned, freezing the UI or a headless renderer for an attacker-chosen duration at negligible cost to the attacker. Exploitation is remote and unauthenticated per the supplied CVSS 4.0 vector, the upstream fix is visible in pull request #2366, and no public exploit code or CISA KEV entry was reported at time of analysis.
An authenticated administrator of a Netcore NR255-V router running firmware 1.5.130703 can crash the device by sending a crafted HTTP request to the route_policy_add.cgi handler that omits the required exit_port parameter, triggering a null pointer dereference (CWE-476) and a denial-of-service condition. The flaw is remotely reachable over the network but sits behind the management interface's high-privilege authentication requirement (CVSS 4.0 PR:H), and it affects availability only — there is no confidentiality or integrity impact. No public exploit code is confirmed in the provided intelligence, the issue is not listed in CISA KEV, and no EPSS score was supplied; the assigned CVSS 4.0 base score is 6.9.
Crafted
Authentication downgrade in the miniOrange JWT Authentication for WP REST APIs WordPress plugin (versions before 4.8.0) lets unauthenticated remote attackers force the plugin to fall back to Basic HTTP authentication by supplying a crafted GET parameter, overriding whatever JWT or API token configuration the site administrator set. Because the vulnerable code path performs no capability check or nonce verification, and the plugin returns distinguishable error codes without rate limiting, attackers can chain the bypass into unthrottled username enumeration and credential guessing against WordPress REST endpoints. Exploitation yields only low confidentiality and integrity impact (CVSS 4.0 base 6.9), and no public exploit code or active exploitation was identified in the supplied intelligence at time of analysis.
Arista EOS devices with password-based authentication enabled can accumulate orphan authentication sessions when a specially crafted password string is submitted to the management login path. Repeatedly sending such passwords exhausts the device's authentication resources, so legitimate administrators are locked out of the switch. Arista discovered the flaw internally and states it has no evidence of malicious exploitation in customer networks; no public exploit code or CISA KEV listing was identified in the supplied intelligence. The vendor-rated CVSS 4.0 score is 6.9 with availability-only impact.
The MCPVault npm package (a lightweight Model Context Protocol server that gives AI agents access to an Obsidian vault) anchored its restricted-directory deny-list to the vault root in versions before 0.11.5, so nested .git, .obsidian and node_modules directories silently pass both isAllowed() and isAllowedForListing(). An attacker who influences which path an AI agent reads — typically through prompt injection in content the agent ingests — can cause nested .git/config files (remote URLs, embedded tokens) or nested Obsidian metadata to be read, and can pollute the listAllTags index with nested node_modules content. No public exploit code or CISA KEV entry was identified at the time of analysis; the vendor shipped the fix in 0.11.5.
Unauthenticated server-side request forgery in AVideo through version 29.0 lets remote attackers drive the server's own HTTP client to arbitrary destinations by supplying crafted Host headers to submitIndex.php or ajax.php. Because the check_site_availability function follows redirects without validating the host, an attacker can map internal hosts and open ports from the outside. No public exploit code or CISA KEV entry was provided in the intelligence feed, so exploitation status is unconfirmed, but the flaw requires no credentials and is trivially automatable.
Opening a crafted .rdp connection file in the FreeRDP xfreerdp X11 client (versions 3.11.0 through 3.30.0) can trigger out-of-bounds heap reads and writes because the client fails to validate MonitorIds array values in xf_detect_monitors. This local, user-interaction-gated flaw requires a victim to open the malicious file and does not affect the RDP server side; publicly available exploit code exists, but there is no confirmation of active exploitation (not in CISA KEV). The vendor has released a patch in FreeRDP 3.31.0, and the CVSS 6.9 rating reflects limited confidentiality and integrity impact with a high availability impact.
Heap buffer overflow in the NGINX HTTP/3 module (ngx_http_v3_module) can crash NGINX worker processes or corrupt a limited amount of heap data when NGINX Plus or NGINX Open Source serves QUIC traffic while linked against OpenSSL 3.5.0 or earlier under certain configurations. The overflow occurs while processing a TLS handshake and is non-deterministic — triggering it is largely outside the attacker's control — so the realistic outcome is opportunistic denial of service (worker restart) plus limited data corruption rather than reliable remote code execution. There is no control plane exposure (data plane only), no public exploit identified at time of analysis, and no CISA KEV listing; EPSS data was not provided.
MISP 2.5.45 and earlier fails to apply its authentication-failure logging throttle on two API auth paths — requests presenting no API key and requests presenting a key of incorrect length — allowing an unauthenticated remote caller to write an unbounded number of durable auth_fail entries into the database. Repeated requests can exhaust storage and degrade or deny availability of the platform (CWE-770). No public exploit code and no CISA KEV entry were present in the supplied intelligence; an upstream fix commit (2bf887433) is available from the vendor.
Adjacent-network attackers can bypass authentication in Oracle Agile PLM 9.3.6 (Application Server component), gaining the ability to read, create, delete, or modify any data the PLM application can reach. The flaw is unauthenticated per its CVSS vector (PR:N) but only reachable from the physical communication segment attached to the host running Agile PLM, and Oracle rates it as difficult to exploit (AC:H) with a 6.8 base score affecting confidentiality and integrity only. No public exploit code or CISA KEV listing was present in the available intelligence, so active exploitation is unconfirmed.
Oracle iRecruitment, the recruitment module of Oracle E-Business Suite, exposes an access-control weakness in its Internal Operations component across versions 12.2.3 through 12.2.15 that lets an attacker on the same physical network segment read, create, modify, or delete recruitment data without authenticating. The flaw is rated difficult to exploit (CVSS 6.8, AV:A/AC:H/PR:N/UI:N) because it demands layer-2 adjacency to the application host and a race-prone or otherwise unreliable attack path, but it delivers high confidentiality and integrity impact with no availability loss. No public exploit code or CISA KEV entry was present in the supplied intelligence, so exploitation is not confirmed in the wild.
Oracle Product Hub, the product master-data component of Oracle E-Business Suite (versions 12.2.3 through 12.2.15), is exposed to a difficult-to-exploit flaw in its Internal Operations component that lets a low-privileged, network-positioned attacker read and tamper with critical product data via HTTP. The CVSS 3.1 base score is 6.8 (AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:N), reflecting high confidentiality and integrity impact but no availability impact, with the high attack-complexity metric indicating exploitation depends on conditions outside the attacker's direct control. There is no public exploit identified at time of analysis and the flaw has not been listed in CISA KEV; Oracle's own advisory at cspusep2026.html is the authoritative remediation source.
Unauthenticated attackers positioned on the adjacent network segment of a host running Oracle Helidon can read and modify critical application data through a flaw in the helidon-integrations-neo4j component. Affected releases are Helidon 3.0.0 through 3.2.20 and 4.0.0 through 4.5.4; Oracle rates this CVSS 3.1 6.8 with an AV:A/AC:H profile, meaning the attacker needs layer-2 proximity and must satisfy non-trivial conditions, but needs no credentials and no user interaction. No public exploit code and no CISA KEV entry were identified at the time of analysis, and no EPSS score was supplied with the intelligence.
Oracle HR Intelligence, a component of Oracle E-Business Suite (versions 12.2.3 through 12.2.15), contains a difficult-to-exploit vulnerability in its Internal Operations component that lets a low-privileged, network-positioned attacker with an authenticated HTTP session tamper with or destroy critical HR data and repeatedly crash the service. Integrity and availability are both fully impacted while confidentiality is not, giving a CVSS 3.1 base score of 6.8 (AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:H/A:H). No public exploit code has been identified at time of analysis, and the issue is not listed in CISA KEV, so the current exposure is driven by unpatched enterprise ERP deployments rather than active exploitation.
Oracle Siebel CRM Deployment (Cloud Gateway component) versions 17.0 through 26.7 contains a hard-to-exploit flaw that lets a low-privileged attacker with HTTP network access read and modify critical CRM data, with Oracle rating both confidentiality and integrity impact as High. The vendor advisory does not disclose the root cause, and no public exploit code, EPSS score, or CISA KEV entry was supplied with this record, so real-world exploitation is unconfirmed. The associated 'Authentication Bypass' tag is not corroborated by the Oracle text, which describes unauthorized data creation, deletion, and modification rather than an explicit credential-check bypass.
The CookieJar client middleware in http4s before 0.23.35 and 1.0.0-M47 stores a response cookie's Domain attribute without verifying that it domain-matches the host that set it, so any malicious or compromised server reached through the same cookie jar can plant a cookie addressed to an unrelated victim domain. When the client later talks to that victim domain, the jar replays the attacker-controlled cookie, enabling session fixation or overwriting of security-sensitive cookies; the flaw has CVSS 6.8 (CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:N/I:H/A:N) with no public exploit code identified at time of analysis, and fixes land in 0.23.35 and 1.0.0-M47.
Http4s client applications that keep a cookie jar can leak stored cookies — including session and authentication cookies — to attacker-controlled hosts because versions prior to 0.23.35 and 1.0.0-M47 decide cookie applicability with unanchored substring checks instead of RFC 6265 domain and path matching. A cookie scoped to example.com is therefore also attached to requests for evilexample.com, api.example.com.attacker.net, or an unrelated path, so any attacker who can influence the outbound URL the application requests receives the cookie and can hijack that outbound session. The flaw is fixed in 0.23.35 and 1.0.0-M47; no public exploit code and no CISA KEV entry were identified at time of analysis.
Improper permission preservation during archive decompression in the HashiCorp go-getter library (1.8.x up to 1.8.8 and 2.x up to 2.2.3) lets a crafted archive cause extracted files to be created with attacker-influenced, elevated permission bits. When the extraction is performed by a privileged process — for example a root-run provisioning, CI, or orchestrator agent — a local actor can leverage the resulting files to gain the privileges of that extracting process, escalating from a low-privileged position to root or service-level access. HashiCorp has released fixes in go-getter 1.8.9 and 2.2.4; the issue is not listed in CISA KEV and no public exploit code has been identified at time of analysis.
Local attackers who already hold high privileges and a valid logon to the host running Oracle Hyperion Financial Management 11.2.26.0.000 can exploit a flaw in the product's Security component to seize full control of the application, with complete loss of confidentiality, integrity and availability. The issue is rated CVSS 6.7 (AV:L/AC:L/PR:H/UI:N) and no public exploit code, EPSS score or CISA KEV entry was present in the supplied intelligence. Because it requires pre-existing privileged local access, this is primarily a post-compromise escalation and insider-abuse concern rather than an internet-facing remote attack.
An out-of-bounds write in the Android kernel GNSS link driver lets a local process that already holds System execution privileges corrupt kernel memory and escalate to kernel-level code execution. The defect (CWE-787) is a missing bounds check in link_load_gnss_image() in link_device.c, requires no user interaction, and is addressed through Google's September 2026 Android/Pixel security bulletin. No public exploit code or confirmed active exploitation has been identified at time of analysis.
Local privilege escalation in the arm-smmu-v3 driver of Google's Android kernel lets code already running with System privileges bypass the ARM SMMUv3 protection mechanism and reach higher-level (kernel/root) execution on affected ARM-based devices, including Pixel. The root cause is a logic error classified as CWE-693 (Protection Mechanism Failure), reachable without any user interaction but only from a local, already-privileged context (CVSS PR:H). No public exploit code and no active exploitation have been identified at time of analysis, so this is a monthly-bulletin hardening fix rather than an emergency patch.
Local privilege escalation in Google Android's GPU driver component allows a process that already runs with System execution privileges to bypass a permission check through a logic error, gaining full read/write/execute impact at a higher privilege level. Only Android devices carrying the affected GPU component are exposed, and exploitation is strictly local with no user interaction, so it cannot be triggered by a remote attacker or a normal unprivileged app on its own. There is no public exploit code and no CISA KEV listing identified at time of analysis; remediation is delivered through the September 2026 Android/Pixel security bulletin.
Local privilege escalation in the Android kernel's arm-smmu-v3 driver (smmu_install_nested_ste) lets a process that already runs with System execution privileges turn a logic error in SMMU stream-table-entry handling into full kernel-level compromise of confidentiality, integrity, and availability. Only devices whose kernel exposes the ARM SMMUv3 driver in a nested-translation configuration are affected, and Google rates it Moderate with a CVSS base of 6.7 (AV:L/PR:H), meaning exploitation presupposes a highly privileged local foothold rather than remote or unauthenticated access. There is no indication of public exploit code and the issue is not listed in CISA KEV; Google's Pixel bulletin of 2026-09-01 ships the fix.
A use-after-free in the ARM SMMU v3 driver (arm-smmu-v3.c) of the Android kernel lets a local attacker who already holds System-level execution privileges corrupt kernel memory and escalate to full kernel/System privilege. The flaw is a logic error in one of several driver functions that leaves a freed object reachable, and no user interaction is needed once the attacker has code running at that privilege level. It is ranked CVSS 6.7 (AV:L/PR:H), there is no indication of active exploitation or public exploit code, and Google published a fix in the September 2026 Pixel security bulletin.
A logic error in the ARM SMMU v3 driver's smmu_detach_dev path allows a caller who already holds System-level privileges on an Android device to bypass SMMU/IOMU memory-permission checks and escalate to kernel (EL1) execution, achieving full compromise of the device. Affected are Google Pixel/Android builds using the arm-smmu-v3 driver prior to the September 2026 Android Security Bulletin patch level. Exploitation is local, requires no user interaction, and no public exploit code or CISA KEV listing was identified at time of analysis, so the practical risk is concentrated on devices that already run privileged or partially compromised code.
A confused-deputy flaw (CWE-441) in the Android platform's message-handler initialization routine, platform_msg_handler_init in default_msg_handlers.c, lets a local process that already runs with System privileges have privileged operations performed on its behalf by a more trusted component, resulting in local privilege escalation with high impact to confidentiality, integrity and availability. Exploitation requires local execution at the System privilege level (CVSS PR:H) and no user interaction, so it is not reachable by an ordinary unprivileged app. There is no public exploit code identified and the issue is not in CISA KEV; it was published through the September 2026 Android/Pixel security bulletin, so risk to most users is conditional on already having a System-level foothold or being part of an exploit chain.
A missing permission check in the FSM key-release routine (FsmReleaseKey in fsm.c) of the Android kernel allows a locally positioned attacker to bypass an access-control verification and escalate privileges on affected devices. The flaw is rated CVSS 6.7 (AV:L/AC:L/PR:H/UI:N) and is classified as CWE-693, Protection Mechanism Failure, because the intended permission gate is simply not enforced on that code path. Exploitation requires the attacker to already hold System-level execution privileges on the device, and no public exploit code or CISA KEV entry has been identified at time of analysis.
Local escalation of privilege in the Android kernel's ARM SMMU v3 driver lets a process that already runs with System privileges reach kernel-level code execution by passing malformed input to the nested device-detach path (smmu_detach_dev_nested in arm-smmu-v3.c). Because the CVSS vector requires PR:H, this is not a boundary an ordinary unprivileged app can cross — it is a second-stage escalation for code that already holds highly privileged, kernel-adjacent access on the device. There is no public exploit identified at time of analysis and the issue is not listed in CISA KEV, so risk is driven by the monthly Android/Pixel kernel patch cycle rather than by observed attacks; no EPSS score was supplied in the input data.