OS command injection in Dell Cloud Disaster Recovery (versions 20.2 and prior) lets an already high-privileged, remote user smuggle special shell characters into a command the product executes on the underlying host, resulting in arbitrary command execution. Dell rates it CVSS 9.1 with a scope change, meaning the injected commands can escape the application's authorization boundary onto the host OS. No public exploit identified at time of analysis and the flaw is not on CISA KEV, so severity is driven by impact rather than confirmed activity.
Remote code execution in Google Cloud BigQuery Data Transfer Service (versions prior to the 2026-05-01 service release) allows an authenticated tenant user to run arbitrary code inside the managed connector container and escalate privileges within the tenant project by supplying crafted parameters in a JDBC connection string routed through the bundled CData JDBC driver. Rated CVSS 4.0 9.4 (Critical) with a scope change to subsequent systems, the flaw was fixed server-side by Google on 1 May 2026 and requires no customer action. There is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Kernel-state confusion in the Linux MPTCP subsystem allows a remote peer to leave stale Multipath TCP Fast Open (MPTFO) state on a passively-opened socket by sending a valid-cookie TCP Fast Open SYN that carries no payload. Because mptcp_fastopen_subflow_synack_set_params() set the is_mptfo flag before verifying that data was actually queued, a data-less TFO SYN triggers a WARN and, if non-fatal, leaves the subflow in an inconsistent state that can later be tripped inside check_fully_established(). The upstream fix marks the subflow as MPTFO only after an SKB is confirmed queued and returns quietly on an empty receive queue. No public exploit identified at time of analysis; the input carries the inflated CVSS 9.4 typical of automated kernel-CVE scoring rather than a demonstrated high-impact primitive.
Unauthenticated arbitrary file upload in Thinking Software Technology's EFence (versions up to and including 1.2.66, DB Ver:56) lets remote attackers upload executable web shell backdoors and run arbitrary code on the underlying server. The flaw requires no authentication or user interaction over the network, yielding full compromise of confidentiality, integrity, and availability. Reported by TWCERT; no public exploit identified at time of analysis, and EPSS exploitation probability is moderate at 0.68% (50th percentile).
Authentication coercion in Veeam ONE lets an unauthenticated network attacker force the Veeam ONE service account to perform outbound SMB authentication to an attacker-chosen destination, exposing the account's NTLM credentials for capture and relay. The flaw (CWE-288, CVSS 4.0 base 9.3) is remotely triggerable with no privileges or user interaction and carries high subsequent-system impact, meaning captured service-account credentials can be relayed to pivot onto other systems. There is no public exploit identified at time of analysis and it is not listed in CISA KEV.
Unauthenticated PHP object injection in the ILIAS e-learning platform allows remote attackers to achieve code execution as the web server user via the Shibboleth back-channel logout endpoint (shib_logout.php). The endpoint runs on an authentication-exempt path and passes every live session row's stored data through a hand-rolled parser that calls unserialize() with no class allowlist, so an attacker who first seeds a serialized object into a session row (via the equally unauthenticated LTI entry point) triggers a destructor-based POP chain that writes attacker-controlled JSON to an attacker-chosen path below the web root. No public exploit has been identified at time of analysis, but the flaw is unauthenticated and network-reachable, and the vendor CVSS 4.0 base score is 9.3 (Critical).
Out-of-bounds read in the RISC-V ZBB-optimized strnlen() routine of the Linux kernel allows the aligned-word loop to load one machine word past the caller-supplied count boundary; when the source is aligned and count is a multiple of the register size, the extra load lands exactly at s+count and faults if that address sits on an unmapped page boundary. The affected code path only runs on RISC-V hardware exposing the Zbb bitmanip extension. This has been fixed upstream (stable commits e697e30f and 5d588c68); no public exploit identified at time of analysis, and the CNA's 9.4 network-vector scoring is inconsistent with an arch-local kernel string-length primitive.
Broken authorization in getgrav/grav-plugin-api before 1.0.18 lets a scoped API key that belongs to a super-admin account perform privileged user-management actions it was never scoped for. Across seven sensitive endpoints the requireNotSuperTarget() guard in UsersController.php checks isSuperAdmin() on the underlying account instead of whether the presented key carries super authority (isSuperWithinScope()), so a deliberately down-scoped key can act against other super-admin accounts-disabling their 2FA, deleting avatars, and minting or deleting their API keys. Reported by VulnCheck; no public exploit identified at time of analysis, and it is not on CISA KEV.
Improper access control on the debug and engineering interfaces of Danfoss iC7-Automation SP, iC7-Marine, and iC7-Hybrid GR3 drive platforms lets attackers read and write internal device values and, more critically, upload and execute unsigned applications, EEPROM data, and firmware through exposed service and software-update mechanisms. Rated CVSS 4.0 9.3, the flaw effectively bypasses code-signing and grants full control over affected industrial drive controllers. There is no public exploit identified at time of analysis, but the missing authentication and firmware-overwrite capability make this a severe integrity and availability risk for operational technology environments.
Cross-tenant data exposure in Google Cloud Vertex AI Search for Commerce (versions prior to 2026-04-27) allowed an authenticated attacker who knew a victim's Google Cloud project number to read from and write to the Cloud Storage buckets used for BigQuery Import staging, because those bucket names were predictably derived rather than randomized. An attacker could reach another tenant's staged catalog data and import error logs, breaching data confidentiality and integrity across project boundaries. There is no public exploit identified at time of analysis, and Google fixed the issue server-side so no customer action is required.
Broken authorization in the Grav API plugin (getgrav/grav-plugin-api) before 1.0.18 lets a holder of a scoped API key view - and potentially edit - a page's security/permission blueprint section beyond the scope their key was granted. The flaw lives in injectSecurityTab() of BlueprintController, which decides editability using raw isSuperAdmin()/hasPermission() checks and never calls scopeAllows(), so the per-key scope cap is silently bypassed. There is no public exploit identified at time of analysis and no KEV listing; the reporter (VulnCheck) explicitly could not confirm the end-to-end write-time impact, so confirmed effect is at least unauthorized disclosure of permission fields.
Uncontrolled recursion in the Linux kernel's s390 vfio-ccw driver lets a local actor on IBM Z (s390) hosts submit a channel program containing an unbounded number of chained CCW/Transfer-in-Channel (TIC) segments, which are processed recursively and can exhaust the kernel stack. Affected are s390 systems using vfio-ccw device passthrough to KVM guests; the impact is host instability and potential kernel memory corruption. No public exploit is identified at time of analysis, it is not in CISA KEV, and no EPSS score was provided.
Host-side memory corruption/information disclosure in the Linux kernel s390 vfio_ccw driver allows a malicious or compromised KVM guest on IBM Z mainframes to exploit a double-fetch race when the driver reads a guest-supplied IDAL (Indirect Data Address List). Because the first IDAW is read separately to size the transfer buffer, a guest that changes that address between reads can desynchronize the buffer size from the actual IDAW set, corrupting or leaking host memory (S:C, C:H/I:H/A:H, CVSS 9.3). No public exploit identified at time of analysis and the issue is not in CISA KEV; the fix (verifying the first IDAW is unchanged between reads) is available upstream via stable kernel commits.
Denial of service in qTox v1.18.4 lets an attacker crash the client through the profile-serialization code in src/persistence/serialize.cpp, classified as uncontrolled resource consumption (CWE-400). The description scopes this to a local attacker requiring victim interaction, and SSVC records a proof-of-concept, so publicly available exploit code exists; it is not listed in CISA KEV and EPSS is low (0.13%, 3rd percentile), indicating no evidence of widespread exploitation. Impact is limited to availability of the desktop client for the affected user, not remote code execution.
Unauthenticated disclosure and modification of runtime configuration in NebulaGraph (vesoft-inc nebula) through version 3.8.0 lets any network-reachable attacker read and rewrite daemon gflags via the built-in web service. Because the service in src/webservice/WebService.cpp binds to all interfaces with no authentication, token, or address restriction, an attacker can harvest sensitive paths (certificate, key, CA, and password-file locations) and disable transport-security flags, redirect logging, and weaken lockout controls without a restart. No public exploit is identified at time of analysis, but the flaw is trivial to trigger and was reported by VulnCheck.
Complete authentication bypass in TarsWeb (the web console for the TARS microservice framework) through 3.0.14 lets a remote unauthenticated attacker act as any account, including an administrator, by sending a forged X-Forwarded-For header naming the loopback address together with a uid query parameter. Because app.js trusts Koa's proxy option blindly and the SSO middleware treats the loopback allowlist match as authorization to adopt the caller-supplied uid, no ticket, cookie, or password is ever checked. No public exploit is identified at the time of analysis, but the technique is trivial and the CVSS 4.0 base score is 9.3 (Critical).
Command injection in Ubiquiti's UID Enterprise Agent lets a high-privileged, network-adjacent actor run arbitrary OS commands on the host device by supplying improperly validated input. The flaw carries a CVSS 9.1 rating driven by a scope change and full compromise of confidentiality, integrity, and availability, though exploitation requires existing high privileges (PR:H). No public exploit has been identified at time of analysis, and it is not listed in CISA KEV.
Command injection in Ubiquiti UniFi OS Server allows an attacker holding high-privilege access and network reachability to execute arbitrary operating-system commands on the underlying host, per Ubiquiti Security Advisory Bulletin 067. The flaw stems from improper input validation (CWE-20) and, with a scope-changed CVSS 9.1, lets a compromised or malicious privileged account break out of the application context to fully control the host. No public exploit identified at time of analysis, and it is not listed in CISA KEV.
Command injection in Ubiquiti UniFi OS Server allows an attacker holding high-privilege (administrative) access on the network-reachable management interface to break out of the application and run arbitrary operating-system commands on the underlying host device. The flaw stems from improper input validation (CWE-20) and carries a vendor-assigned CVSS of 9.1, driven largely by a scope change from application to host. No public exploit identified at time of analysis, and it is not listed in CISA KEV; the only source is Ubiquiti's own Security Advisory Bulletin 067.
Authenticated command injection in Ubiquiti's UniFi Network Application (the controller software managing UniFi network gear) allows an attacker holding high-privilege access on the network to inject and execute operating-system commands on an adopted device managed by the controller. Because the injected command runs on a downstream device rather than the controller itself, the flaw crosses a trust boundary (CVSS scope change) and yields full confidentiality, integrity, and availability impact on the target device. No public exploit identified at time of analysis, and the issue is not listed in CISA KEV.
Remote code execution in Google langfun before 0.1.2 arises because the default lf.query Python protocol evaluates model-generated Python expressions with no sandbox, letting remote unauthenticated attackers run arbitrary code in the host application by supplying crafted prompt inputs that steer the model into emitting executable code. Because langfun embeds directly in LLM-driven Python applications, any deployment that pipes untrusted prompt content through lf.query is exposed. There is no public exploit identified at time of analysis and it is not in CISA KEV, but the CVSS 4.0 base score of 9.2 (CRITICAL) reflects full confidentiality, integrity, and availability impact.
Privilege escalation in Ubiquiti's UniFi Network Application allows an already-authenticated, high-privileged user with network access to gain elevated rights within the application through an improper access control (CWE-284) flaw. Ubiquiti self-reported the issue in Security Advisory Bulletin 067 and tags it as an authentication bypass; there is no public exploit identified at time of analysis and it is not listed in CISA KEV. The vendor rates it CVSS 9.1, driven largely by an assumed scope change, though the practical bar is raised by the requirement for pre-existing high privileges.
Information disclosure in the OpenRISC OR1200 processor soft-core (commit 83ac6b) arises from a mismatch between the RTL source and the synthesized netlist, meaning the fabricated/synthesized hardware behaves differently than the verified design and can leak confidential state or misbehave. The flaw was surfaced by the academic SynFuzz RTL-vs-netlist fuzzing research (arXiv:2504.18812) rather than by field exploitation; no public exploit identified at time of analysis and EPSS is low (0.15%, 5th percentile). The NVD CVSS of 9.1 (C:H/A:H) is far more alarming than the vague one-line description supports, so the score should be treated skeptically.
Server-side request forgery in JetLinks Community 2.11 lets attackers abuse the device metadata import endpoint (/device/instance/{productId}/property-metadata/import) to coerce the server into issuing arbitrary outbound requests. Because the platform sits at the IoT control layer, this can be used to reach internal-only services, cloud metadata endpoints, and other network-segment resources normally shielded from the internet. A public write-up/POC exists (GitHub) and SSVC rates the flaw as automatable with total technical impact, though EPSS remains low (0.14%) and it is not on CISA KEV, so no public exploit identified at time of analysis as active exploitation.
Server-side request forgery in Zyplayer-Doc version 1.0.0 and earlier lets remote unauthenticated attackers coerce the server into issuing arbitrary outbound requests through the WikiPageWebService.download() function. Because the wiki page download feature fetches attacker-controlled URLs without validation, an attacker can reach internal-only services, cloud metadata endpoints, and other back-end systems the server can see. Publicly available exploit code exists (a GitHub write-up), CVSS is rated 9.1, and SSVC flags the flaw as automatable with total technical impact, though it is not listed in CISA KEV.
Hard-coded credentials in TÜBİTAK BİLGEM's Liderahenk Linux endpoint management platform (all versions before 3.5.5) let a remote, unauthenticated attacker log in using built-in default/common username-password pairs, yielding an authentication bypass with high confidentiality and integrity impact. Because Liderahenk centrally administers fleets of Pardus/Linux clients, valid access to its management interface exposes managed-endpoint policies and configuration. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but the flaw was reported by Turkey's national CERT (TR-CERT).
Authentication bypass in Ubiquiti UniFi OS lets a network-adjacent attacker circumvent login on affected devices by injecting CRLF sequences, granting full control over the appliance under certain conditions. It affects a broad range of UniFi OS platforms including Cloud Gateways, Dream Machines/Routers, Cloud Keys, and Enterprise NVR/NAS hardware. Reported by Ubiquiti with a CVSS of 9.0; there is no public exploit identified at time of analysis and it is not listed in CISA KEV, though the high scope-changing impact makes it a priority to patch.
Privilege escalation in Ubiquiti's UniFi Connect Display Cast Pro lets a network-adjacent attacker abuse an improper access-control flaw to gain elevated rights on the device under certain conditions. The issue was reported by Ubiquiti and disclosed in Security Advisory Bulletin 067; it is tagged as an authentication bypass. No public exploit identified at time of analysis, and it is not listed in CISA KEV, but the vendor-assigned CVSS of 9.0 marks it as critical.
Privilege escalation in Ubiquiti UniFi OS affects a broad range of gateway, NVR, storage, and Cloud Key appliances running the UniFi OS platform, where leftover Active Debug Code (CWE-489) lets an authenticated low-privilege actor with network access elevate their privileges on the device. The flaw is vendor-reported via Ubiquiti Security Advisory Bulletin 067 and carries a 9.0 CVSS driven by a security-scope change and full confidentiality, integrity, and availability impact. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but the wide device footprint makes it a notable patch priority.
Missing authentication on Spring Cloud Config Server's /monitor endpoint lets remote attackers submit unvalidated webhook requests, forcing configuration-refresh events without any credentials. It affects Spring Cloud Config 3.x through 3.1.14, 4.0.0-4.2.8, 4.3.0-4.3.4, and 5.0.0-5.0.4, and is exploitable only when the spring-cloud-config-monitor component is enabled. No public exploit identified at time of analysis; EPSS is low (0.28%, 20th percentile) and CISA SSVC rates exploitation as none and technical impact as only partial, tempering the vendor's 9.8 rating.