An out-of-bounds stack write in the VobSub subtitle decoder (libsub_vob) of the Amlogic DroidLogic SubTitleService allows remote attackers to corrupt memory on Nexbox A95X (Amlogic S905X) TV boxes and other devices built from the Amlogic Android SDK. A crafted SET_CONTR command carrying an attacker-controlled alpha value drives the change_data index past the end of the fixed four-element PixelDatas array inside vob_fill_pixel(), zeroing adjacent stack bytes; the payload can arrive over the unauthenticated tcp/10100 socket or by getting the device to parse a malicious .sub file. No authentication and no user interaction are required, but the assessed impact is limited to silent stack corruption or a process crash rather than code execution, and the tcp/10100 service is typically LAN-exposed rather than internet-facing. A public PoC reference exists; it is not verified as working exploit code, and no confirmed active exploitation has been reported.
An infinite loop in the get_vob_spu() function of the libsub_vob VobSub subtitle decoder, part of Amlogic's DroidLogic SubTitleService shipped on Amlogic S905X-based devices such as the Nexbox A95X TV box, lets remote network attackers reachable on the LAN or local users loading a crafted file stall the subtitle decode thread at 100% CPU indefinitely, producing a denial of service. Two trigger paths are documented: sending a crafted AMLU data block with mismatched length fields to the unauthenticated tcp/10100 listener, or loading a malicious VobSub (.sub) file. No authentication or user interaction is required (CVSS 4.0 base 6.9 Medium; assessed CVSS 3.1 AV:N/AC:L/PR:N/UI:N with availability-only low impact), and only a public PoC reference exists - it is not verified as working exploit code, and no active exploitation has been confirmed. Exposure is bounded by the fact that tcp/10100 is a local service typically reachable only from the LAN, with no confidentiality or integrity impact and no code execution.
Code injection in dayrui XunRuiCMS through 4.7.4 allows remote, unauthenticated attackers to manipulate the dirname argument processed by str_replace in the Mobile Site Directory component (/zzmob/index.php), with low impact on confidentiality, integrity and availability. The CVSS 4.0 base score is 6.9 (medium) with E:P; a public PoC reference exists but is not verified as working exploit code, and no CISA KEV entry is associated. Exploitation requires the zzmob mobile-site feature to be present and reachable.
Server-side code injection in the email-template rendering path of dayrui XunRuiCMS up to 4.7.4 (_get_tpl_content in dayrui/Fcms/Library/Notice.php) lets an attacker who can influence rendered template content have injected code evaluated by PHP, giving limited confidentiality, integrity and availability impact. The vendor was contacted and did not respond, and no fixed version is listed. A public proof-of-concept reference exists (a GitHub disclosure report against 4.7.2) but is not verified as working exploit code, and there is no KEV or confirmed active exploitation; our independent assessment rates this medium severity (CVSS 3.1 AV:N/AC:L/PR:L/UI:N, C:L/I:L/A:L) and the trigger path is not described in the available sources, so whether template editing requires an authenticated administrative session - a strong limiting factor - should be verified per deployment.
SQL injection in the Member Module of dayrui XunRuiCMS through 4.7.4 lets remote attackers manipulate the 'catid' argument handled by the View function in dayrui/Fcms/Core/View.php. The CVSS vector indicates no authentication or user interaction is required, so requests can be sent directly against a vulnerable instance, though the member endpoint must be reachable and the precise how the parameter reaches the query is not detailed in available sources. Impact is bounded to Low confidentiality, integrity, and availability per the assessed vector; exploit code is referenced publicly but is not verified as working code, and no confirmed active exploitation or CISA KEV listing is associated with this CVE.
Unauthenticated SQL injection in mathurvishal's CloudClassroom-PHP-Project lets remote attackers manipulate the eid parameter of viewquery.php to alter the SQL query the application executes, exposing and potentially modifying data stored by the classroom application. The flaw is remotely reachable without credentials or user interaction (CVSS:3.1/AV:N/AC:L/PR:N/UI:N), but impact is scored as limited confidentiality and integrity loss rather than full compromise; the product is a niche individual open-source project, so real-world exposure depends on whether an instance is network-accessible. A public PoC reference exists; it is not verified as working exploit code, the issue is not on CISA KEV, and the vendor did not respond to disclosure and ships a rolling release, so no version-specific fix information is available.
CloudClassroom-PHP-Project exposes protected student records through studentdetails.php because the script continues executing after an access-control redirect (CWE-698). The endpoint is network-reachable with low complexity and no authentication or user interaction per the assessed vector, but impact is limited to low-grade confidentiality loss and the project has a tiny rolling-release footprint. A public PoC reference exists; it is not verified as working exploit code, and no vendor-released fixed version is identified.
A use-after-free in the P2P session teardown handler of VStarcam O-KAM SuperIPCam 1.3.3.32 (SuperIPCam.exe, function FUN_005a2d44) allows remote, unauthenticated attackers who can reach the camera's P2P connectivity feature to corrupt memory during session teardown, producing a partial loss of confidentiality, integrity and availability consistent with a denial-of-service crash rather than confirmed code execution. Only version 1.3.3.32 is identified as affected, and the vendor was contacted about the disclosure but did not respond. A public PoC reference exists (a GitHub gist linked from the VulDB entry); it is not verified as working exploit code, and there is no CISA KEV listing or other confirmation of active exploitation.
A stack-based buffer overflow (CWE-121) exists in the Windows client software VStarcam O-KAM Client 1.3.3.32, specifically inside the exported/imported function FUN_10003EA0 of the native library DevDll_925.dll. The vendor was contacted but did not respond and no vendor-released patch has been identified at time of analysis; the affected version is the only one confirmed in the CPE data (cpe:2.3:a:vstarcam:o-kam_client), so exposure is limited to users running that build. The issue is remotely reachable according to the CVSS 4.0 vector (AV:N/PR:N/UI:N, score 6.9; our assessed CVSS 3.1 vector scores it 6.5 with Low confidentiality, integrity and availability impact), and a public PoC reference exists in a GitHub gist; it is not verified as working exploit code and there is no CISA KEV confirmation of active exploitation. The exact feature, protocol, or code path that delivers attacker-controlled data into FUN_10003EA0 is not described in the available sources.
Path traversal in the Aklivity Zilla http-filesystem binding lets remote unauthenticated attackers who can reach the binding manipulate the params argument so that Path.resolve in FileSystemServerFactory escapes the intended filesystem root, exposing files the Zilla process can read. All releases up to and including 1.3.4, 2.4.7 and 2.5.0-rc2 are affected, but only deployments that actually configure and expose the http-filesystem binding are reachable; the assessed impact is confidentiality-only, bounded by the process's host read permissions. A public PoC reference exists (a write-up in a GitHub CVE repository); it is not verified as working exploit code, no CISA KEV listing confirms active exploitation, and vendor fixes are available in 2.5.0-rc3.
Unauthenticated order-forgery in the PayHere Payment Gateway WordPress plugin, versions 2.4.0 through 2.4.5, allows an attacker to mark any order as paid without making a payment. The plugin fails to verify the authenticity of incoming payment notifications when the merchant secret is left unset - the default configuration - so a forged confirmation is accepted; sites that have configured a merchant secret are not exploitable. No authentication or user interaction is required (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:N, 5.3), impact is confined to order-status integrity with no data disclosure or availability effect, there is no CISA KEV entry, and only a public PoC reference exists that has not been verified as working exploit code; the vulnerability is tracked as a this_month priority item.
Stored cross-site scripting in Munshi (by Petpooja) before the build released on 2026-07-13 allows authenticated users to persist malicious markup via Purchase Inquiry item descriptions or thread messages at /purchase-inquiry/thread/{thread-id}, which the application assigns to innerHTML while rendering before text cleaning completes. The payload fires in the browser of any other user who opens or reopens the affected thread, giving script execution and full-page defacement that re-triggers on every revisit. Exploitation requires a valid Munshi account and a victim viewing that specific thread, and a public PoC reference exists; it is not verified as working exploit code. There is no CISA KEV listing, and this is assessed as a moderate-risk, authenticated stored XSS rather than a top-priority emergency, with patching aligned to this month's cycle.
Stored cross-site scripting in GoFound through 1.4.1 lets unauthenticated attackers plant HTML/script by submitting document text to the /api/index endpoint, with no credentials required (CVSS:4.0 PR:N). The payload stays inert in the index until an administrator runs a dashboard search that matches the poisoned document, at which point searcher/engine.go returns the stored content unescaped and dashboard.vue renders it via v-html, executing in the admin's browser session. Exploitation is therefore interaction-gated and scoped to the administrator's browser rather than the server; a public PoC reference exists but is not verified as working exploit code, and there is no CISA KEV confirmation of active exploitation. The issue is rated 5.1 and prioritised as "this_month" rather than an emergency.
PandaWiki 3.65.0 through 3.87.5 contains a stored cross-site scripting flaw in the unauthenticated /share/v1/common/file/upload endpoint, where an attacker can upload HTML content under an image extension. The stored payload executes when an administrator is lured to /static-file/<key>?a=.pdf, because the query string defeats nginx attachment handling and causes the HTML to render inline, enabling theft of the panda_wiki_token JWT and session takeover. Exploitation requires administrator interaction and a valid admin session; a public PoC reference exists but is not verified as working exploit code, and the provided data do not confirm active exploitation.
Path traversal in the AWS Serverless MCP Server (awslabs.aws-serverless-mcp-server) before v0.2.1 allows an actor who can influence the MCP agent's serverless project-name input to read the contents of JSON files outside the intended deployment status directory, because the deployment status component does not confine the resolved pathname to that directory. When the server is additionally run with the non-default --enable-write flag, the same input can create or overwrite JSON files outside the directory, clobbering existing files with serverless deployment metadata. The CVSS 3.1 vector (AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:L/A:N) reflects no privileges required but mandatory interaction, and no public exploit was identified at time of analysis; the vendor states there is no workaround and that upgrading to v0.2.1 is the fix.
Server-side request forgery in newspaper4k through 0.9.6 lets an unauthenticated remote attacker who can influence a page processed by an integrating application cause the Image Extractor to fetch image or meta-refresh URLs pointing at loopback, link-local, private, or internal hosts. The impact is limited to confidentiality - limited internal response content may be exposed - with no integrity or availability effect, and the supplied CVSS 4.0 score is 6.9. The vendor fixed the issue in 0.9.6.1; no public exploit identified at time of analysis.
Repository-internal file disclosure in Otter Wiki 2.20.1 through 2.26.1 lets anyone who can edit a wiki page read files stored inside the application's own repository - most critically .git/config, which frequently holds remote URLs with embedded credentials - by passing an absolute src (for example /.git/config) to the DataTable embedding, which resolves the path and renders the file contents as a table. Under the default configuration anonymous visitors can edit pages, so the CVSS vectors carry PR:N and no account is needed; instances that disable anonymous editing or restrict edit rights require a valid account with edit permission instead. Impact is bounded to reading repository-internal files (confidentiality only, no write or denial-of-service effect), scoring 6.9 under CVSS 4.0 and 5.3 under the assessed CVSS 3.1 vector; there is no CISA KEV listing and no public exploit identified at time of analysis, while a vendor fix is available in 2.26.2.
Time-based blind SQL injection in Control Web Panel (CWP) releases prior to webpanel_version 1.30 allows an attacker who can reach an internal, loopback-bound API route - originally /user_panel_dashboard/, later /valid_token/ - to inject the 'user' parameter into a query executed against the oauth_clients table over a MySQL connection opened as root, then read arbitrary database contents by inferring data bit-by-bit. Because the route is bound to loopback rather than exposed publicly, reaching it requires local access to the server or an SSRF/relay primitive in another service; once that reachability exists, no authentication (PR:N) and no user interaction are required. The vendor changelog records a patch, and no public exploit was identified at time of analysis; the slow, inference-based extraction is the main reason this ranks as a this-month priority rather than an emergency despite a high confidentiality impact.
The Eclipse ThreadX Module Manager fails to enforce thread ownership in its thread-control dispatchers, so a loaded user-mode, memory-protected module can suspend, resume, terminate, reset, re-prioritise and abort the waits of application and other-module threads it does not own. The flaw affects ThreadX 6.0.1 through 6.5.2.202603 deployments that use the Module Manager with memory-protected modules; a malicious or compromised module can cause denial of service and manipulate scheduling, but gains no code execution or memory access. Exploitation requires local (on-device) control of an already-loaded memory-protected module, and the issue is fixed in ThreadX 6.5.2.202603b; no public exploit code has been identified at time of analysis.
Heap out-of-bounds write in L-SMASH through 2.14.5 lets a crafted MP4 or MOV file corrupt heap memory inside applications that use the library to demux media, typically ending in a crash. The flaw lives in the ISO-BMFF box parsers isom_read_hdlr() and isom_read_dref_entry() in core/read.c, which trust a declared hdlr or dref 'url ' box size larger than the bytes actually remaining in the stream, stalling the stream position and driving the write past the allocated buffer. The assessed vector is local (AV:L) with user interaction (UI:R), no authentication, no confidentiality impact, limited integrity impact and high availability impact (CVSS 3.1 6.9 equivalent); an upstream fix commit exists, but no tagged patched release or working exploit code has been independently confirmed, and the issue is not in CISA KEV.
Out-of-bounds write in rapidyaml (ryml) through 0.16.0 lets an attacker who can supply a YAML document containing consecutive or excessive anchors/tags overflow the parser's fixed two-element annotation array in ParseEngine::_add_annotation(), corrupting stack memory and crashing the host process in release builds. The CVSS 4.0 vector (AV:L/AC:L/AT:N/PR:N/UI:P, with VA:H and VI:L) means unauthenticated attackers at the local vector need only get an application to process the crafted document, and the assessed consequence is high availability impact with limited integrity effect rather than confidentiality loss. No CISA KEV entry exists and no public exploit code is confirmed at time of analysis, though a third-party technical writeup describing the parser annotation overflow is referenced; vuln.today priority is this_month, i.e. a moderate, non-emergency issue.
Improper access control in Outline before 1.12.0 leaves previously shared documents readable by unauthenticated users after the owner withdraws them: the public share loader in server/commands/shareLoader.ts retrieves the document through the 'withDrafts' scope and never enforces publishedAt IS NOT NULL, so an existing share link keeps serving the full text. Anyone holding the share URL can pull the withdrawn content through POST /api/shares.info, POST /api/documents.info with shareId, GET /s/<share-id> and GET /s/<share-id>.md without authentication, and while the document stays unpublished the shares.revoke endpoint returns 404 for both the share owner and workspace administrators, so the exposure cannot be closed through the normal UI. The vendor advisory includes a detailed public proof-of-concept walkthrough (not verified working exploit code); the issue was reported by VulnCheck and is fixed in 1.12.0 (commit d015a32459b5f5c5a8ef67aeb81742eeab430c52).
SQL injection in OwnTone Server through 29.3 lets remote attackers read arbitrary media library data through the MPD emulation listener on TCP port 6600. The mpd_sticker_find() handler interpolates the attacker-supplied URI argument directly into a SQLite LIKE filter, so a crafted 'sticker find song' command escapes the intended virtual-path restriction and enumerates every media file along with its filesystem path and metadata. No authentication or user interaction is required, but exploitation depends on the MPD listener being reachable - typically it is bound to a trusted home or LAN network rather than the public internet - and the impact is confined to information disclosure (integrity and availability are unaffected). No public exploit code has been identified at time of analysis and the issue is not in CISA KEV; the vendor has published an upstream fix commit.
Heap memory exhaustion in QuickJS-NG versions before 0.16.0 lets JavaScript executed by an embedding host drain the process one BigInt allocation at a time: each Atomics.store() call on a BigInt64Array or BigUint64Array whose backing buffer is detached (ab.transfer()) or shrunk (ab.resize(0)) during value coercion leaks the converted BigInt instead of freeing it. An attacker able to run scripts inside a QuickJS-NG host - server-side script runners, plugin sandboxes, IoT firmware - can loop that pattern to cause denial of service in long-running processes; there is no confidentiality or integrity impact and no code execution. Vendor CVSS 4.0 is 6.9 with availability impact Low only, the priority label is 'this_month' rather than emergency, and no public exploit code or CISA KEV entry was identified at time of analysis; the fix ships in release 0.16.0.
Out-of-bounds read in Netpbm's tgatoppm converter allows a crafted color-mapped Targa file to crash the tool or leak adjacent memory into the output image. Affected builds are Netpbm before 10.86.51, 11.x before 11.02.30, and 11.03 before 11.15.08; exploitation is local and requires the victim or an automated workflow to run tgatoppm on the malicious file. No public exploit identified at time of analysis, and a vendor patch is available; this is assessed as a moderate-priority issue, not an emergency.
Network-AI's A2AServer fails open on authentication: the `secret` option is optional with no default and `startServer()` performs no fail-closed startup check, so an instance started without a secret accepts unauthenticated JSON-RPC `tasks/send`, `tasks/get` and `tasks/cancel` calls from anyone who can reach the listener. The npm package `network-ai` is affected through version 5.15.4 (vendor fix: 5.15.5); the default 127.0.0.1 bind limits exposure, but the A2A protocol is designed for inter-agent network communication, so operators who bind a reachable interface-or attackers using a co-located/container-network process or DNS rebinding against loopback-can submit attacker-controlled task text to the developer-supplied executor, read task results and cancel tasks. Rated moderate (CVSS 6.9, CWE-306), the issue is not in CISA KEV and no confirmed active exploitation exists, though the GitHub advisory publishes a curl-based proof-of-concept reference; the downstream impact beyond task records and executor output depends entirely on what the configured executor does with the supplied text.
Code execution can be achieved in Dell OpenManage Enterprise versions prior to 4.8.0 by a high-privileged attacker with adjacent-network access who can exploit improper verification of cryptographic signatures. The CVSS 3.1 vector (AV:A/AC:H/PR:H) and CWE-347 indicate that exploitation requires high privileges, adjacent-network positioning, and high-complexity conditions; the 'Jwt Attack' tag suggests a forged or unverified JWT may be involved, but the exact signed object and trigger are not described in the available sources. This is a moderate-severity issue rather than a top-tier emergency, with no confirmed active exploitation (not in CISA KEV) and no public exploit identified at time of analysis.
Android's USB service can be induced to enable USB data access while the device is in a strong-authentication-required state (for example a locked or post-boot locked device where USB data should remain restricted), because of a logic error in onStrongAuthRequiredChanged within UsbService.java. The result is a physical privilege escalation: an attacker with the device in hand and a USB connection gains access to data that should have been blocked, with no authentication and no user interaction. It is rated 6.8 (AV:P/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H), affecting Android 15, 16, 16-qpr2 and 17 per the November 2026 Android security bulletin; no public exploit is identified in the available data and CISA's SSVC assessment records exploitation as none and automatable as no, so realistic risk is confined to device-theft, evil-maid and forensic-extraction scenarios rather than remote or scalable attack.
A permission bypass in Google Android 14, 15, 16, 16-qpr2 and 17 allows an attacker with physical access to the device to escalate privileges, with no authentication or user interaction required once that physical contact is achieved. Severity is moderate (CVSS 6.8, CWE-693 protection mechanism failure), and the practical exposure is confined to lost, stolen, seized or unattended-device scenarios rather than remote attacks. No public exploit is identified in the available data and EPSS is very low (0.13%, 3rd percentile); the Android security bulletin of 2026-11-01 is the reference, and no fixed version is listed in the supplied evidence.
Format-string handling in psqlODBC before 18.00.0006 can crash applications or disclose client memory when an updatable cursor reads a table without OIDs that has an auto-increment int4 column whose server-reported name contains printf directives. The driver uses that column name as a format string while building the WHERE clause for positioned SELECT, UPDATE, and DELETE statements; exploitation requires a server not trusted under current SSL settings or a database user able to create a table the application reads, plus the non-default cursor/table/column alignment. This is a moderate, not emergency, priority, and no public exploit was identified at time of analysis; the assessed vector is AV:N/AC:H/PR:N/UI:R with high confidentiality and availability impact.
Eclipse ThreadX 6.0.1 through 6.5.2.202603 can be left with its Module Manager permanently deadlocked: when a module preamble declares a data size or a start/stop/callback stack size that overflows the allocation arithmetic, the loaders return TXM_MODULE_MATH_OVERFLOW without releasing the Module Manager mutex, so every later module load, start, stop or unload from any other thread blocks until the system is reset. Affected are systems that actually use the Module Manager and load images through txm_module_manager_in_place_load(), txm_module_manager_memory_load(), txm_module_manager_absolute_load(), txm_module_manager_absolute_load_extended() or txm_module_manager_file_load(); the vector is local with PR:L (CVSS-B 6.8), so the actor must be able to supply a crafted module image. Impact is availability-only - no out-of-bounds read or write occurs and confidentiality and integrity are unaffected - and no public exploit has been identified at time of analysis; the vendor released a fix in 6.5.2.202603b.
Denial of service in the Kvazaar HEVC command-line encoder through 2.3.2 lets an attacker hang the encoder by supplying a crafted .y4m input whose header is truncated and lacks the terminating newline (0x0A); getc() then returns EOF without being checked, so the parse loop in read_header() never makes progress and the process spins indefinitely on a full CPU core. The flaw is local in nature - it requires an operator to run the encoder on attacker-supplied input (UI:P), needs no privileges (PR:N), and produces no confidentiality or integrity impact, only high availability impact. A public PoC reference exists in the form of a published writeup; it is not verified as working exploit code, and there is no KEV confirmation of active exploitation. The only complete fix per the available evidence is updating to code containing the upstream parser rewrite.
Denial of service in The Sleuth Kit through 4.15.0 occurs when a crafted HFS+ image containing an LZVN inline com.apple.decmpfs attribute with a huge uncompressed_size value is parsed, causing an allocation failure and a NULL pointer write in decmpfs_decompress_lzvn_attr(). The flaw affects forensic analysts and automated pipelines that load untrusted HFS+ images; it is a local, user-assisted crash-only issue with no code execution or data compromise. No CISA KEV entry is present and no public exploit identified at time of analysis.
MediaTek's Download Agent (DA), the bootloader/flash-mode download component present on a broad set of MediaTek chipsets, contains an out-of-bounds write that allows an attacker with physical access to the device to escalate privileges, with no credentials, no additional execution privileges and no user interaction required once that physical access exists. The affected scope covers the chipsets enumerated in the EUVD list, including MT6983, MT6985, MT6989, MT6990 and MT6991. Exploitation is not automatable, no public exploit has been identified at time of analysis, and the fix is referenced by Patch ID ALPS11468968 in MediaTek's November 2026 product security bulletin.
Local privilege escalation in Google Android 17 (addressed through the Wear OS security bulletin dated 2026-11-01) is caused by an out-of-bounds write in is_valid_header() in timed_cache_item.c, where a missing bounds check allows memory beyond the intended buffer to be overwritten. Exploitation is local only (CVSS AV:L) and requires the attacker to already hold System execution privileges (PR:H), with no user interaction needed, so the flaw is a post-compromise consolidation primitive rather than an initial-access vector. EPSS is 0.15% (4th percentile), no public exploit has been identified at time of analysis, and the CVE is not in CISA KEV; vuln.today classifies it as this_month priority.
Improper quoting of shell variables in konflux-ci deployment scripts allows unintended command substitution, so crafted arguments can run arbitrary commands with the privileges of the deployment runner. The exposure is local and requires high privileges (CVSS:3.1/AV:L/PR:H/UI:N/S:U/C:H/I:H/A:H, 6.7): an attacker must already be able to supply arguments to the affected scripts inside the deployment context, and impact is bounded by the runner's own permissions. No public exploit code has been identified at time of analysis, and this is not a remotely or unauthenticated exploitable issue.
Elevation of privilege in Dell Alienware Command Center (AWCC) releases prior to 6.14.54.0 lets a local, non-administrative user gain higher-integrity execution on the host, with the vendor advisory also citing arbitrary code execution and denial of service as possible outcomes. Dell scores this at CVSS 6.7 with a local attack vector, high attack complexity, low privileges required and user interaction (CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:U/C:H/I:H/A:H), so an attacker must already hold a foothold on the machine and must induce the user to do something before the least-privilege weakness (CWE-272) can be exercised. No public exploit code or in-the-wild exploitation has been identified at time of analysis, and the vendor recommends updating to AWCC 6.14.54.0 per DSA-2026-429.
Dell OpenManage Enterprise versions prior to 4.8.0 fail to properly verify cryptographic signatures, which can let an attacker who already holds high-privileged local access on the appliance gain filesystem access. The CVSS 3.1 vector (AV:L/AC:H/PR:H/UI:N) means exploitation requires pre-existing administrative privileges on the host and carries high attack complexity; the trigger mechanism and the specific feature or token involved are not described in the available sources, though the issue is tagged as a JWT-related signature attack. There is no public exploit identified at time of analysis and no CISA KEV listing, and effective remediation is available in OpenManage Enterprise 4.8.0 per Dell advisory DSA-2026-467.
Local privilege escalation in the Modem component of a broad range of MediaTek chipsets lets a process that already holds the System privilege bypass permission assignments and gain higher-level capabilities (C:H/I:H/A:H). The CVSS 3.1 score of 6.7 with vector AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H is honest about the main constraint: exploitation is local and requires the attacker to have already obtained System privilege, so this is a second-stage escalation rather than a remotely reachable entry point, and no user interaction is needed. Affected silicon spans the MT27xx, MT68xx, MT69xx, MT86xx, MT87xx and MT88xx families listed in MediaTek's November 2026 product security bulletin; no public exploit was identified at time of analysis and KEV/EPSS signals were not provided.
Out-of-bounds write in the MediaTek Modem component lets a local attacker who already holds System privilege escalate further on affected MediaTek chipsets. The missing bounds check (CWE-787) can corrupt memory and produce high confidentiality, integrity, and availability impact, but only after a high-privileged local foothold and with no user interaction. No public exploit was identified at time of analysis, and MediaTek tracks the fix as Patch ID MOLY01320076 in its November 2026 product security bulletin.
Local privilege escalation exists in the Modem component of a broad range of MediaTek chipsets, where a missing permission check (CWE-280) lets code running with System privilege reach higher-integrity operations, with high impact on confidentiality, integrity and availability (CVSS 6.7, AV:L/AC:L/PR:H/UI:N/S:U). The prerequisite is explicit in the vendor description: an attacker must already hold the System privilege on the device, so this is a post-compromise escalation step rather than an entry point, and it is reachable locally with no user interaction and no network vector. No public exploit code was identified at time of analysis and there is no CISA KEV listing; MediaTek has published a fix under patch ID MOLY01790985 in its November 2026 product security bulletin, and the affected hardware spans dozens of chipsets including MT6761, MT6765, MT6768, MT6833, MT6877, MT6895, MT6983, MT8768 and MT8788E.
Out-of-bounds write in MediaTek's NeuroPilot AI/ML (APU) component allows a local attacker who already holds System privilege to escalate further on devices running affected MediaTek chipsets, with high impact to confidentiality, integrity and availability. Exploitation is local (AV:L), requires high privileges (PR:H) and needs no user interaction (UI:N), so it cannot be used for initial access or remote compromise - it is a post-compromise escalation primitive. No public exploit code has been identified at time of analysis (SSVC exploitation: none, CISA KEV: not listed); MediaTek has released patch ALPS11263078 in its November 2026 product security bulletin.
Local privilege escalation in MediaTek chipset USB handling allows an attacker who already holds System privilege to exploit a missing bounds check, corrupt memory, and escalate privileges on affected devices. The issue affects the MediaTek chipsets enumerated in EUVD-2026-115048, including MT6769, MT6833, MT6877, MT6895, MT6983, MT6991 and others listed in the advisory. No public exploit was identified at time of analysis.
Double free in the USB driver shipped with several MediaTek chipsets (MT6993, MT8668, MT8188, MT8518S, MT8532, MT8189) corrupts kernel memory and can be converted into a local privilege escalation by an attacker who already holds System privilege on the device; no user interaction is required (CVSS 6.7, AV:L/PR:H/UI:N). MediaTek has assigned fix ALPS11411924 (issue MSV-9748) and published it in the November 2026 product security bulletin. No public exploit code was identified at time of analysis and CISA SSVC records exploitation as none, so the practical risk profile is a post-compromise escalation step rather than an initial-access or remote vector.
A use-after-free memory corruption flaw in the USB component of several MediaTek chipset models (MT8189, MT8668, MT6881, MT8532, MT6993, MT8518S, MT6858, MT8188) can be triggered locally by an attacker who already holds System privilege, allowing further escalation of privileges. Because the pre-existing System privilege requirement is the dominant limiting factor, this is a conditional, lower-tier priority rather than an urgent emergency; exploitation requires local access at System level, and no user interaction is needed. No public exploit code or active exploitation has been identified at time of analysis, and the precise USB operation that triggers the use-after-free is not described in the available sources.
Local privilege escalation in the MediaTek afe audio front-end driver affects devices built on a set of MediaTek chipsets, including MT6985, MT6886, MT6993, MT6895, MT6897, MT6991, MT6989 and MT6789. An attacker who already holds System privilege can trigger a missing bounds check that leads to an out-of-bounds write, gaining high confidentiality, integrity and availability impact without user interaction; the issue is local only and does not provide an initial foothold. No public exploit was identified at time of analysis, and CISA SSVC records exploitation as none and automatable as no.
Use-after-free in the MediaTek audio front-end (afe) driver on affected MediaTek chipsets allows a local attacker who already holds System privilege to escalate to a higher privilege level, with high confidentiality, integrity, and availability impact and no user interaction (CVSS 3.1: 6.7, AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H). The issue affects Android devices built on a broad set of MediaTek chipsets and has a vendor patch identified by ALPS11431394; no public exploit identified at time of analysis.
This is a genuine but constrained privilege-escalation issue rather than a top-tier emergency. Local privilege escalation in the MediaTek afe audio front-end driver affects Android devices using the listed MediaTek chipsets; an attacker who already holds System privilege can trigger memory corruption to gain higher-level code execution with high confidentiality, integrity, and availability impact. No user interaction is required, but the PR:H precondition confines this to a post-compromise escalation step, and no public exploit identified at time of analysis.
Local privilege escalation in MediaTek's afe component on affected chipsets is caused by a double-free memory-corruption flaw that a local attacker can trigger after already holding System privilege. Successful exploitation can compromise confidentiality, integrity, and availability at higher privilege, with no user interaction required. No public exploit was identified at the time of analysis; MediaTek assigns Patch ID ALPS11431412 in its November 2026 product security bulletin.
MediaTek adsp contains an out-of-bounds write caused by an integer overflow that can allow a local actor who already holds System privilege to escalate privileges further on affected MediaTek chipsets, with no user interaction required. The issue is rated 6.7 (CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H) and affects a broad set of chipsets including MT8390, MT6989, MT8196, and MT6991. MediaTek has published patch ALPS11431442; no public exploit was identified at time of analysis.