In the Airoha Bluetooth audio SDK, there is a possible way to pair Bluetooth audio device without user consent.
In the Airoha Bluetooth audio SDK, there is a possible way to pair Bluetooth audio device without user consent. This could lead to remote escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.
Unauthenticated RCE in Kaltura mwEmbed HTML5 Player
Kaltura's HTML5 player library (mwEmbed/html5lib) contains an unauthenticated remote code execution vulnerability (CVE-2026-19912) caused by unsafe PHP deserialization combined with unsanitized filesystem path construction in mwEmbedLoader.php. A remote attacker controls the ServiceUrl parameter, whose response is passed to unserialize(), and supplies a crafted uiconf_id so that fields of the resulting object are written to an attacker-influenced cache path without path validation, allowing arbitrary files to be written to web-accessible locations and code execution as the webserver user; the related advisory headline also notes the flaws could allow reading files. The flaw is rated CVSS 9.8 (critical), requiring no privileges or user interaction over the network. Any deployment serving an affected html5lib v2.x release, including v2.45, v2.103 and earlier, is affected. There is no known public proof-of-concept, the issue is not in CISA's KEV catalog, and EPSS estimates only a 0.6% probability of exploitation within 30 days (46th percentile), indicating no confirmed exploitation yet.
· Kaltura mwEmbed HTML5 player (html5lib) v2.45, v2.103 and earlier, and other v2.x releases exposing the vulnerable mwEmbedLoader.php endpointlarge
Heap-Based Buffer Overflow in Konami Metal Gear Online 3 Enables Code Execution
CVE-2026-19874 is a critical heap-based buffer overflow (CWE-122) in the lobby-data handling of Konami's Metal Gear Online 3: the code trusts the "kick_num" field for the count of kicked-player identifiers (supplied as "kicked_id_%i" keys) without validating it against the 16-entry buffer sized for the game's maximum match size. An attacker who can supply crafted lobby data, such as a lobby host or participant, can set "kick_num" above 16, causing the parser to write past the end of the kicked-ID buffer and into adjacent memory containing Steam callback handler structures, overwriting function pointers and callback arguments. This can lead to control-flow hijacking and potentially arbitrary code execution within the game process; the CVSS 9.1 score reflects network reachability with no privileges or user interaction required. Any player running the affected title who joins or hosts lobbies with untrusted peers is exposed (the described code path touches Steam callback structures, indicating the Steam-integrated client); no fixed version information is included in the available data. There is currently no public proof-of-concept, no known in-the-wild exploitation, and the issue is not in CISA's KEV; EPSS estimates a 0.7% probability of exploitation within 30 days (52nd percentile).
· Konami Metal Gear Online 3niche
Privileged local BIOS code execution in AMI Aptio V (incomplete input validation)
AMI's Aptio V UEFI BIOS firmware contains an input-validation flaw classified as CWE-184 (Incomplete List of Disallowed Inputs), meaning a BIOS code path fails to reject all inputs it is supposed to disallow. Triggering it requires local access to the machine, and the CVSS 4.0 metrics show the attacker must already hold high privileges (typically OS-level administrator), with no user interaction needed but elevated attack prerequisites that must be met. Successful exploitation yields arbitrary code execution in the firmware context with high impact on confidentiality, integrity and availability on the affected system and, per the scoring, potentially on subsequent systems - significant for BIOS because firmware-level compromise can persist across OS reinstalls. Any workstation, server or motherboard shipped with AMI Aptio V firmware by an OEM is in scope. No public proof-of-concept, CISA KEV listing, or known in-the-wild exploitation has been reported (EPSS 0.1%); a separate related AMI firmware issue (VU#718077, Secure Boot bypass via a UEFI Shell embedded in SPI flash) is also documented.
· AMI (American Megatrends International) Aptio V (AptioV) UEFI BIOS firmwaremass
Secure Boot bypass via embedded UEFI Shell in UEFI BIOS firmware (CVE-2026-6485)
CVE-2026-6485 (CWE-489, active code left in production firmware) describes a UEFI Shell module embedded in the SPI Flash of affected UEFI BIOS images that can be used to bypass Secure Boot. An attacker with local access and high privileges (CVSS AV:L/PR:H) — for example a local administrator or someone with physical access — can run shell commands or execute startup scripts through the embedded shell, launching code that Secure Boot does not verify. The result is a break in the Secure Boot chain of trust, allowing unsigned or attacker-controlled code to run at boot (e.g., to install bootkit- or firmware-level persistence), which is why the CVSS scope is 'changed' (S:C) with high impact to confidentiality, integrity, and availability. Affected systems are those whose UEFI BIOS ships with this embedded UEFI Shell module in SPI Flash; the issue is tracked in CERT/CC VU#718077, but the exact vendor, product list, and version ranges are not specified in the available data. There is no known exploitation at this time: no public proof-of-concept exists, the issue is not in CISA KEV, and EPSS estimates only a 0.1% chance of exploitation in the next 30 days.
· UEFI BIOS firmware with a UEFI Shell module embedded in SPI Flash
Pre-Consent Remote Python File Write in Hugging Face Transformers
CVE-2026-80047 is a flaw in Hugging Face Transformers (versions 4.49.0 through 5.8.1) in which GenerativePreTrainedModel.load_custom_generate() writes a remote Python file (custom_generate/generate.py) from a model repository to the local ~/.cache/huggingface/modules directory before performing the required trust_remote_code consent check. The unconditional file write in dynamic_module_utils.py occurs even when the user declines the trust prompt, inverting the consent-first model enforced by AutoConfig, AutoModel, and AutoTokenizer. Execution of the code is correctly gated, but the write is persistent, so attacker-controlled code remains on disk and can later be executed via stale-cache collisions during otherwise trusted model loads. Affected users are developers, CI pipelines, and applications running affected Transformers versions that load custom generate implementations from remote repositories. There is currently no public proof-of-concept, no known exploitation (EPSS 0.1%), and the issue is not in CISA KEV; it was assigned by CERT/CC under VU#456290.
· Hugging Face Transformers >= 4.49.0 and <= 5.8.1niche
Unauthenticated Local File Disclosure in Kaltura HTML5 Player (mwEmbed)
The Kaltura HTML5 player (mwEmbed/html5lib) fails to properly validate the ServiceUrl parameter in mwEmbedLoader.php, which is used as the base URL for a backend request and accepts non-HTTP schemes such as file://. An unauthenticated remote attacker can supply a file:// URL, and when the backend request raises an exception or error, the server deserializes the response and reflects its raw contents back to the client in an error message, disclosing arbitrary files readable by the web server user. An attacker gains read access to internal server files, which may include configuration files or credentials, with confidentiality impact only per the CVSS score. Any deployment exposing the vulnerable mwEmbedLoader.php endpoint is affected, including sites running html5lib v2.45, v2.103 and earlier, and other v2.x releases. No public proof-of-concept, KEV listing, or confirmed in-the-wild exploitation is known; EPSS is low at 0.4%, though related reporting indicates additional unpatched mwEmbed flaws, possibly including code execution, may exist.
· Kaltura html5lib / mwEmbed HTML5 player (mwEmbedLoader.php) v2.103 and earlier (including v2.45) and other v2.x releases exposing the vulnerable mwEmbedLoader.php endpoint; no fixed release identified in available reportlarge
Secure Boot Bypass via UEFI Shell in Cisco UCS Servers and UCS-based Appliances
CVE-2026-20293 is a flaw in the Unified Extensible Firmware Interface (UEFI) Shell implementation of Cisco UCS Servers and UCS-based appliances that lets an attacker bypass UEFI Secure Boot validation and run unauthorized software. It is triggered because memory-write commands remain available in the UEFI Shell even while Secure Boot is enabled: an attacker selects the UEFI Shell boot option at boot time and uses shell commands to modify UEFI memory variables and overwrite Secure Boot-related memory values. A successful exploit manipulates the preboot environment and allows execution of untrusted software that would normally be rejected by Secure Boot. Exploitation requires either physical access (no credentials needed) or valid credentials for an account with the user or admin role on the affected system, so practical exposure is limited to systems where an attacker has local access. No exploitation in the wild, public proof-of-concept, or KEV listing is known, and EPSS puts 30-day exploitation probability at about 0.1%.
· Cisco UCS Servers (UEFI Shell implementation) · Cisco UCS-based appliances (UEFI Shell implementation)large
A vulnerability was determined in AOMEI Backupper up to 8.3.0.
A vulnerability was determined in AOMEI Backupper up to 8.3.0. Impacted is an unknown function in the library amwrtdrv.sys of the component Kernel Driver. Executing a manipulation can lead to improper access controls. The attack needs to be launched locally. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way.
—
SSRF in ONLYOFFICE ownCloud Integration plugin 9.12
CVE-2026-84282 is a server-side request forgery (SSRF) in the ONLYOFFICE ownCloud Integration plugin version 9.12, where the /apps/onlyoffice/ajax/settings/address endpoint does not sufficiently validate the user-supplied Document Server URL before the ownCloud server makes outbound connections. To trigger it, an authenticated administrator supplies a manipulated 'document server' parameter, causing the server to send requests to attacker-chosen destinations, including localhost and internal network hosts. An attacker gains internal network reconnaissance and can perform TCP port scanning of the ownCloud host's local network by differentiating server responses. Only ownCloud deployments running this integration plugin are affected, and exploitation requires administrative access, so an attacker would typically need a compromised or untrusted admin account. There is no known public proof-of-concept, the flaw is not in CISA's KEV catalog, and EPSS puts 30-day exploitation probability at 0.2% (6th percentile).
· ONLYOFFICE ownCloud Integration plugin 9.12moderate