SQL injection in MuuCMF T6 v1.9.4.20260115 lets an unauthenticated remote attacker inject arbitrary SQL through the keyword parameter of the /index/controller/Search.php endpoint, exposing the full database and enabling forged administrative access. Because the underlying database and application run with sufficient privileges, the flaw can be escalated to write malicious files to disk and potentially achieve remote code execution. Publicly available exploit code exists (write-up by thinhneee), but the flaw is not on CISA KEV and EPSS is very low at 0.03% (10th percentile).
Reflected cross-site scripting in Mercury docuForm (Managed Print Services) v11.11c enables authenticated attackers to execute arbitrary JavaScript in victim browsers when users click malicious links. The vulnerability exists in dfm-menu_markeralerts.php due to unfiltered variable handling. CVSS 7.3 (High) reflects network-accessible attack requiring low-privilege authentication and user interaction. No public exploit confirmed, but proof-of-concept code published by ZeroBreach GmbH demonstrates feasibility. EPSS data unavailable; not in CISA KEV, indicating no confirmed widespread exploitation at time of analysis.
Reflected XSS in docuForm Managed Print Services 11.11c enables authenticated attackers to execute malicious JavaScript in victim browsers via crafted links. The dfm-menu_orderopt.php component fails to sanitize input parameters, allowing session hijacking and account takeover when low-privilege users click attacker-controlled URLs. EPSS data unavailable; no CISA KEV listing indicates exploitation remains opportunistic rather than widespread. Public exploit code exists via GitHub Gist, lowering barrier to attack.
Reflected cross-site scripting in Mercury Managed Print Services docuForm v11.11c allows authenticated attackers with low privileges to execute arbitrary JavaScript in victim browsers via the acc-menu_pricess.php component. Public proof-of-concept exploit code exists on GitHub (ZeroBreach-GmbH). EPSS data not available, not listed in CISA KEV. Attack requires user interaction (phishing/social engineering to click malicious link), limiting automated exploitation but enabling credential theft and session hijacking for authenticated users.
Reflected cross-site scripting in Mercury docuForm v11.11c allows authenticated attackers to execute arbitrary JavaScript in victim browsers via crafted payloads to dfm-menu_alerts.php. Attack requires low complexity and user interaction (CVSS:3.1/AV:N/AC:L/PR:L/UI:R). Public proof-of-concept exists on GitHub (ZeroBreach-GmbH gist), but no CISA KEV listing indicates targeted or low-volume exploitation. EPSS data not available, but the authentication requirement and user interaction dependency reduce attack surface compared to stored XSS.
Remote denial-of-service and limited information disclosure in dnsmasq's extract_addresses() function allows attackers controlling or able to inject DNS responses to crash the daemon by sending a malformed DNS record whose declared rdlen pushes extract_name() past the record boundary, triggering a heap out-of-bounds read. The flaw affects dnsmasq 2.93 and downstream packages from Red Hat, SUSE, Ubuntu, and Pi-hole FTL. No public exploit identified at time of analysis, EPSS exploitation probability is 0.03% (9th percentile), and CISA SSVC rates exploitation as 'none' with only partial technical impact.
Heap buffer overflow in dnsmasq's extract_name() function allows remote attackers to write out-of-bounds on the heap, enabling DNS cache poisoning that redirects lookups to attacker-controlled IP addresses or causes denial of service. The flaw stems from a bigname namebuffer sized for wire-form domain names (MAXDNAME) instead of the larger escaped internal representation (MAXDNAME*2 + 1). EPSS is low (0.03%, 9th percentile) with no public exploit identified at time of analysis, but multiple major distributions (Red Hat, SUSE, Ubuntu) have shipped patches reflecting widespread downstream exposure.
Account-lockout bypass in WSO2 Identity Server (6.0.0, 6.1.0, 7.0.0, 7.1.0) and the Carbon MagicLink Authenticator Module allows accounts that have been administratively or automatically locked to still authenticate through the Magic Link and Passkey (FIDO2) flows, because those paths do not validate account state. An attacker in possession of a valid magic link or passkey for a locked account can reach the application and its sensitive data despite the lock, and the lockout control intended to stop further login attempts is rendered ineffective. There is no public exploit identified at time of analysis, EPSS is negligible (0.03%), and the flaw is not in CISA KEV.
Alien::FreeImage versions through 1.001 for Perl contains several vulnerable libraries. Rated high severity (CVSS 7.3), this vulnerability is remotely exploitable, no authentication required, low attack complexity. No vendor patch available.
Out-of-bounds write in Apple's file parsing component across iOS, iPadOS, and macOS enables remote code execution or denial of service via maliciously crafted files with no user interaction required. Exploitation probability is extremely low (EPSS 0.02%, 6th percentile) with no public exploit identified at time of analysis, despite the critical CVSS 7.3 score and network-based attack vector. Vendor patches available for all affected platforms (iOS/iPadOS 18.7.9, 26.5; macOS Sonoma 14.8.7, Sequoia 15.7.7, Tahoe 26.5). The CVSS vector indicating AV:N/PR:N/UI:N suggests automatic exploitation without user interaction, which contradicts the description's 'parsing a file' language - verify whether this requires user action to open/download the file or if background processes parse untrusted files automatically.
Remote code execution in QuickJS-NG 0.12.1 allows unauthenticated network attackers to execute arbitrary code through the js_mapped_arguments_mark function. The vulnerability enables attackers to achieve code injection with low confidentiality, integrity, and availability impact via network-accessible JavaScript engine exploitation. EPSS score of 0.02% (5th percentile) indicates very low predicted exploitation probability, and no active exploitation has been publicly reported at time of analysis.
Arbitrary code execution in CosyVoice (FunAudioLLM's multilingual text-to-speech toolkit) through commit 6e01309 allows an attacker to run code on a victim's machine via a malicious PyTorch .pt file. The vulnerability lives in the make_parquet_list.py data-preparation tool, which calls torch.load() without weights_only=True, so any .pt file it ingests can carry a malicious pickle payload that executes when the victim processes an attacker-supplied data directory. Reported by MITRE with a low EPSS (0.02%, 5th percentile) and no public exploit identified at time of analysis; this is a poisoned-input/supply-chain style flaw rather than a remotely reachable network service.
Arbitrary code execution in the Dao-AILab flash-attention training framework (all revisions up to and including commit e724e2588cbe754beb97cf7c011b5e7e34119e62) arises because its checkpoint-loading paths call PyTorch's torch.load() without weights_only=True, permitting pickle-based deserialization of attacker-controlled objects. Any user who warmstarts or evaluates a model from an untrusted checkpoint file triggers code execution on their host. No public exploit is identified at time of analysis, EPSS is negligible (0.02%), and SSVC records exploitation as 'none' - this is a latent supply-chain/ML-artifact risk rather than an actively exploited one.
Arbitrary code execution in CosyVoice's gRPC server (through commit 6e01309e) occurs because the speech-synthesis model is loaded via torch.load() without weights_only=True, deserializing untrusted pickle data. Any operator who starts the gRPC server against an attacker-supplied model directory executes embedded Python code during initialization. No public exploit has been identified and it is not in CISA KEV; EPSS is negligible at 0.02% (5th percentile), consistent with a file-delivery attack rather than mass remote exploitation.
Arbitrary code execution in CosyVoice (FunAudioLLM's open-source multilingual text-to-speech toolkit) arises because its average_model.py utility calls PyTorch's torch.load() on epoch_*.pt checkpoint files without weights_only=True, allowing pickle-based deserialization of attacker-controlled objects. Any user who runs the model-averaging tool against a directory containing a malicious checkpoint executes attacker code on their machine. There is no public exploit identified at time of analysis and EPSS is very low (0.02%), but the CWE-502 pattern is trivially weaponizable and SSVC marks it automatable.
Arbitrary code execution in the flash-attention project's training script (up to commit e724e2588cbe754beb97cf7c011b5e7e34119e62) allows an attacker to run Python code by supplying a malicious Hydra configuration. The script registers Python's eval() as a Hydra resolver named 'eval', so any config value using the ${eval:...} syntax is executed when a victim launches training with that configuration. No public exploit has been identified and EPSS probability is very low (0.02%), but SSVC rates the flaw as automatable with partial technical impact.
Out-of-bounds read in Apple operating systems allows malicious applications to crash the system or leak kernel memory across iOS/iPadOS 26.5, macOS Tahoe 26.5, tvOS 26.5, and watchOS 26.5. The vulnerability requires local application execution but no user interaction, enabling information disclosure and denial-of-service attacks. Despite high CVSS 7.3 scoring, the EPSS probability is very low (0.02%, 5th percentile), indicating minimal observed exploitation activity. Vendor-released patches are available for all affected platforms.
Stored cross-site scripting in MantisBT (Composer package mantisbt/mantisbt, versions <= 2.28.1) allows any authenticated user to inject arbitrary HTML/JavaScript through the 'font_family' account preference, which is then rendered into the CSS/head of every MantisBT page. Because the account preference update accepted an arbitrary font string without validating it against the allowed font list, the payload persists and executes whenever the affected page context is loaded. There is no public exploit identified at time of analysis and it is not in CISA KEV; EPSS is very low (0.05%), but chained with the companion CSP-bypass advisory GHSA-9c3j-xm6v-j7j3 the flaw can escalate to full account takeover.
Broken access control in MantisBT's programmatic file APIs lets any authenticated REPORTER-level user (access level 25) download attachments from private bugnotes created by higher-privileged DEVELOPER, MANAGER, or ADMIN users, exposing internal logs, screenshots, and patches. The flaw affects the REST endpoint GET /api/rest/issues/{id}/files and the SOAP mc_issue_attachment_get endpoint, while the web UI is unaffected because it filters through bugnote_get_all_visible_bugnotes(). There is no public exploit identified at time of analysis, EPSS is very low (0.04%), and the issue is not on CISA KEV.
Symbolic link path traversal in pgAdmin 4 File Manager allows authenticated users to write arbitrary files on the server filesystem. Attackers with valid credentials can plant symlinks in their storage directory pointing outside it, bypassing access controls to overwrite critical system files or application configurations with pgAdmin process privileges. The vulnerability combines CWE-61 (symlink following) with a time-of-check-time-of-use race condition. Affects all pgAdmin 4 versions before 9.15. No active exploitation confirmed (not in CISA KEV), but exploit is straightforward for authenticated attackers given the detailed fix description published by PostgreSQL project.
Server-Side Request Forgery in MLflow allows authenticated users to force the MLflow backend to send HTTP requests to arbitrary URLs, including internal services and cloud metadata endpoints (e.g., AWS EC2 metadata at 169.254.169.254). Affects MLflow versions prior to 3.9.0. The webhook creation endpoint accepts unvalidated user-controlled URLs that are later used in HTTP POST requests, enabling cloud credential theft, internal network reconnaissance, and data exfiltration. Vendor-released patch available in MLflow 3.9.0, confirmed by GitHub commit 64aa0ab. No active exploitation confirmed (not in CISA KEV), but publicly disclosed with detailed technical analysis from huntr.com.
Authenticated users in pgAdmin 4 before version 9.15 can read arbitrary server-side files or trigger server-side request forgery (SSRF) attacks via unvalidated LLM API configuration endpoints. The vulnerabilities exist in the chat and model-list endpoints where user-supplied api_key_file and api_url parameters are passed directly to LLM provider clients without sanitization, enabling attackers to exfiltrate sensitive files (database credentials, private keys) readable by the pgAdmin process or coerce internal requests to cloud metadata services and private infrastructure.
Insecure Direct Object Reference (IDOR) in Open WebUI allows authenticated users with read-only channel access to modify or delete any message in those channels, bypassing frontend ownership controls through direct API calls. The vulnerability affects the channels feature in Open WebUI versions ≤0.6.18, exploiting missing message ownership validation in backend FastAPI endpoints despite correct frontend access control implementation. Publicly available exploit code exists with detailed proof-of-concept demonstrating privilege escalation from read to write/delete permissions. Vendor-released patch available in version 0.6.19.
Memory-corruption in Apple's operating-system file-processing components (iOS/iPadOS before 18.7.9, macOS Sequoia before 15.7.7, macOS Tahoe before 26.5) lets a remote attacker cause a denial-of-service crash or leak fragments of process memory when a victim opens a maliciously crafted file. Rated CVSS 7.1, it is unauthenticated but requires user interaction to open the file. There is no public exploit identified at time of analysis, EPSS is negligible (0.02%), and it is not on CISA KEV.
Denial of service in the barebox bootloader (all versions up to and including 2026.04.0) lets an attacker on the same broadcast domain crash the device by sending a crafted DHCP Offer or ACK packet that omits the 0xff end marker, triggering an out-of-bounds read in the dhcp_message_type() option parser. Exploitation is unauthenticated but requires adjacent network access, and impact is limited to availability (a crash) rather than code execution or data theft. No public exploit is identified at time of analysis and the EPSS probability is negligible (0.01%, 1st percentile).
User equipment (UE) downlink traffic can be redirected to attacker-controlled radios in Ella Core (5G AMF software) versions prior to 1.10.0. A malicious radio with a valid NG Setup connection can forge PDUSessionResourceSetupResponse messages using arbitrary AMF-UE-NGAP-IDs, causing Ella Core to create GTP tunnels that misdirect victim UE downlink packets to the attacker's radio. This enables traffic interception and denial of service against targeted UEs. The vulnerability stems from missing validation that NGAP messages arrive on the correct SCTP association for the UE context. No public exploit identified at time of analysis, and EPSS data not available. Vendor-released patch: version 1.10.0.
Commit and tag signature confusion in go-git (the pure-Go implementation of Git) lets a crafted Git object carry a signature that validates while its effective metadata differs from what was actually signed. Because go-git parses ambiguous or malformed object headers differently from canonical Git, and because its signing/verification path operates over a reconstructed representation rather than the raw object bytes, an attacker with contributor access can create commits whose displayed author/committer/metadata diverges from the signed payload - undermining downstream trust decisions (the issue was originally surfaced through sigstore/gitsign). Affected releases are go-git v5 before 5.19.0 and v6 alpha builds 6.0.0-alpha.1 through 6.0.0-alpha.2. There is no public exploit identified at time of analysis, it is not listed in CISA KEV, and no EPSS score was provided in the source data.
Path traversal in Audiobookshelf's podcast creation endpoint allows attackers to write files outside the intended library directory by supplying a crafted file path. Affected versions are all releases prior to 2.32.2 (per CVE) or 2.33.2 (per EUVD - see confidence notes), with the vulnerable code located in server/controllers/PodcastController.js. No public exploit code has been identified and CISA KEV has not flagged active exploitation, but SSVC marks the attack as automatable, meaning scripted mass-scanning is feasible once a reliable exploit exists.
Reflected HTML injection in MantisBT's tag update page (tag_update_page.php) arises from the application echoing the HTTP Referer header into an HTML hidden input without proper escaping. All versions through 2.28.1 are affected. In the most exploitable scenario, an attacker poisons a caching proxy by submitting a crafted Referer value, causing downstream users who receive the cached response to execute attacker-controlled JavaScript - a web-cache-deception-style XSS path. No public exploit identified at time of analysis, though SSVC classifies exploitation status as poc; EPSS is low at 0.04% (14th percentile).
Integer coercion error in bettercap's zerogod IPP service allows remote attackers to trigger information disclosure via malformed chunked HTTP requests. Versions up to 2.41.5 are affected. The vulnerability exists in the ippReadChunkedBody function where integer type conversions lack proper bounds validation, enabling attackers to cause memory access violations that leak sensitive data. Publicly available exploit code exists, and while CVSS 2.9 indicates low severity, the attack requires high complexity and has limited impact; however, the vulnerability is confirmed remotely exploitable without authentication.
Verbose error messages in WeGIA versions prior to 3.7.0 expose database-related details through the unauthenticated PHP endpoint `atendido/familiar_docfamiliar.php`, enabling remote attackers to map backend infrastructure and identify further attack vectors. The CVSS 4.0 score of 6.9 reflects network-accessible, zero-authentication exploitation with partial confidentiality impact and no integrity or availability consequence. A proof-of-concept has been identified per SSVC classification, though EPSS sits at just 0.04% (13th percentile), indicating limited broad exploitation interest; no active exploitation has been confirmed via CISA KEV.
pgAdmin 4 before version 9.15 allows unauthenticated attackers to bypass account lockout and perform unbounded password-guessing attacks against INTERNAL authentication accounts by exploiting Flask-Security's default /login endpoint, which does not enforce the locked column that the custom /authenticate/login view relies on for brute-force protection. The vulnerability affects only accounts using pgAdmin's INTERNAL authentication source; LDAP, OAuth2, Kerberos, and Webserver authentication methods are not vulnerable because they do not use local passwords.
Infinite loop denial-of-service in barebox's ext4 directory parser allows an attacker who can supply a crafted filesystem image to permanently hang the boot process of affected embedded devices. Versions prior to 2026.04.0 are affected via CWE-835 (unreachable loop exit) in ext4fs_iterate_dir() when it encounters a directory entry with a direntlen of zero. No public exploit has been identified and CISA has not added this to KEV; however, the impact of a permanently hung bootloader is a complete availability loss for the affected device.
Out-of-bounds read vulnerabilities in barebox's ext4 extent parsing allow an attacker who can supply a malformed ext4 filesystem image - via USB, SD card, or network boot - to crash the bootloader during startup, preventing a device from completing its boot sequence. All barebox versions prior to v2026.04.0 fail to validate the `eh_entries` field in ext4 extent headers against actual buffer capacity in `fs/ext4/ext4_common.c`, enabling heap out-of-bounds reads that can additionally redirect parsing to arbitrary disk offsets. No public exploit has been identified at time of analysis; EPSS (0.01%, 2nd percentile) and SSVC (exploitation: none, automatable: no) both confirm low near-term exploitation probability, though the physical-media attack surface makes this relevant to embedded device supply chain and physical security programs.
Summarize versions through 0.14.1 create daemon configuration files with world-readable permissions, allowing local attackers with user-level access to read bearer tokens and API credentials from ~/.summarize/daemon.json. The vulnerability enables unauthorized daemon access or recovery of sensitive provider API keys through insecure file permissions on Unix-like systems.
Stored Cross-Site Scripting in WeGIA versions prior to 3.7.3 allows authenticated high-privilege users to inject malicious JavaScript into the Processo de Aceitação page that executes when any user accesses the vulnerable endpoint, enabling session hijacking and account takeover. The vulnerability is confined to high-privilege authenticated access (PR:H) with no user interaction required on the victim side, though scope is changed (S:C) indicating cross-context impact. No public exploit code or active exploitation has been identified.
Stored cross-site scripting (XSS) in WeGIA versions prior to 3.7.3 allows authenticated high-privilege users to inject malicious JavaScript into the Etapas de um Processo page, which executes when other users access the page, enabling session hijacking and account takeover. The vulnerability requires high-privilege authentication (PR:H) but affects all users who subsequently visit the injected page, with no public exploit identified at time of analysis.
Insufficient session expiration in Vaultwarden prior to 1.35.5 allows an attacker holding a previously obtained refresh token to maintain persistent vault access even after the legitimate account owner performs security-remediation actions. Operations that rotate the internal security_stamp - including password change, key rotation, KDF change, email change, org admin password reset, and emergency access takeover - fail to invalidate outstanding refresh tokens, leaving stolen sessions alive. A proof-of-concept exists per SSVC data, though EPSS at 0.03% (8th percentile) indicates exploitation remains uncommon and non-automated, consistent with the need for pre-existing token access.
Path traversal in Crabbox <0.9.0 allows local attackers to delete or overwrite arbitrary files via malicious .crabbox.yaml configuration. When a user executes Crabbox commands with a crafted workspace configuration containing directory traversal sequences (e.g., '../../../'), the Islo provider performs rm -rf and mkdir -p on attacker-controlled paths outside /workspace. Patch available in v0.9.0 (commit 6b07193). No KEV listing or public POC identified, but exploitation requires only user interaction (opening/running a malicious project), not authentication or special privileges.
Path traversal in tookie-osint (all versions prior to 4.1fix) allows an attacker who controls the tool's username input to redirect scan output files to arbitrary filesystem paths accessible to the invoking user. The write_txt, write_csv, and write_json helpers in modules/modules.py construct output filenames via unsanitized f-string interpolation of the -u CLI flag or lines from a -U batch usernames file, enabling directory-separator sequences to escape the intended output directory. A proof-of-concept exists per SSVC assessment; however, EPSS sits at 0.01% (3rd percentile) and the vulnerability is absent from the CISA KEV catalog, consistent with the niche footprint of this tool and the local execution prerequisite.
Improper privilege management in Dell ECS 3.8.1.0-3.8.1.7 and ObjectScale prior to 4.3.0.0 allows high-privileged local attackers to escalate privileges and gain full system access, affecting confidentiality, integrity, and availability. No public exploit code or active exploitation has been identified at the time of analysis.
Remote authenticated command injection in Tenda AC6 router firmware version 15.03.06.23 allows authenticated attackers to execute arbitrary OS commands via manipulation of the wl2g.public.country or wl5g.public.country parameters in the /goform/WifiApScan endpoint. The vulnerability affects the httpd component's formWifiApScan function and has publicly available exploit code, presenting moderate risk to affected deployments.
Command injection in GPT-Pilot's Executor.run() method enables arbitrary shell command execution through the interactive command confirmation workflow, affecting all versions through commit 0819827ce20346ef5f25b3fe29293cb448840565 (2025-09-03). An actor with access to the confirmation prompt - locally or via a network-exposed deployment - can substitute malicious shell commands that are passed unsanitized to asyncio.create_subprocess_shell(), executing with the full privileges of the GPT-Pilot process. No active exploitation has been confirmed (not in CISA KEV), and EPSS at 0.14% (33rd percentile) reflects low current exploitation activity, though the RCE impact class warrants attention in multi-user or network-exposed deployments.
Broken authorization logic across two sequential sharing API calls in Outline collaborative documentation (prior to 1.7.0) allows any authenticated low-privilege user to publish arbitrary documents as publicly accessible shares, exposing content to unauthenticated users without possessing legitimate share rights over the target document. The flaw exploits a permission gap in shares.create (which skips share-permission validation when published=false) combined with a disjunctive OR policy in shares.update that can be satisfied by unrelated collection-level share permissions. No public exploit code repository has been identified in the provided references, though SSVC classifies exploitation evidence at POC level; the vulnerability is not listed in CISA KEV, and EPSS sits at 0.04% (12th percentile), indicating low current real-world exploitation.
Streamlink versions 8.3.0 and earlier allow arbitrary local file read via unvalidated URI schemes in HLS and DASH parsers. A remote attacker hosting a malicious `.m3u8` or `.mpd` manifest can trick streamlink clients into reading local files (such as SSH keys, AWS credentials, or `/etc/passwd`) and writing their contents to the output stream. Requires user interaction (clicking/running streamlink against attacker-controlled playlist), but no authentication needed. CVSS 6.5 reflects the information disclosure risk; affected deployments with automated processing or cloud sync pose material exfiltration risk.
Information leakage in Apple operating systems allows remote attackers to extract sensitive data by crafting and hosting malicious websites that users visit. The vulnerability affects iOS 18.7.8 and earlier, iPadOS 18.7.8 and earlier, macOS Sequoia 15.7.6 and earlier, macOS Sonoma 14.8.6 and earlier, macOS Tahoe 26.4 and earlier, tvOS 26.4 and earlier, visionOS 26.4 and earlier, and watchOS 26.4 and earlier. Exploitation requires user interaction (UI:R) to visit a malicious website but does not require authentication, with an EPSS score of 0.03 percent indicating low real-world exploitation probability despite the information disclosure impact.
Memory corruption in Apple's WebKit web content processing engine causes an unexpected process crash (denial-of-service) across iOS, iPadOS, macOS, tvOS, visionOS, and watchOS when a victim processes attacker-controlled web content. The CVSS vector AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H confirms network-reachable, unauthenticated exploitation limited to availability impact within the affected process - no code execution or data exfiltration is indicated. No public exploit code has been identified at time of analysis, and EPSS scoring at 0.03% (10th percentile) combined with SSVC Exploitation status of 'none' signal low current exploitation pressure.
Authenticated users with download permissions in Audiobookshelf prior to 2.32.2 can download files from libraries they do not have access to by directly specifying item IDs in the GET /api/libraries/:id/download endpoint, bypassing library access controls. An attacker with valid credentials and access to any single library can exfiltrate complete file contents from restricted libraries, including those explicitly denied to them.
HTTP::Tiny versions before 0.093 for Perl fail to validate carriage return and line feed (CRLF) characters in HTTP request lines and header values, allowing attackers who control input URLs or headers to inject additional HTTP headers and smuggle requests to upstream servers. Remote unauthenticated attackers can exploit this via crafted URLs passed to webhook or URL fetch endpoints, achieving limited information disclosure and integrity compromise. EPSS score of 0.03% (percentile 7%) indicates low practical exploitation probability despite network-vector accessibility.
Improper authorization in Open WebUI versions up to 0.8.5 allows authenticated users to modify messages in standard channels by exploiting a read-permission check that should require write permission. The vulnerability affects the POST /api/v1/channels/{channel_id}/messages/{message_id}/update endpoint when access_control is set to None, enabling message tampering without ownership verification. Patch available in version 0.8.6.