Vulnerability database
Every flaw the U.S. government’s cybersecurity agency (CISA) has confirmed attackers are using, plus the new WordPress plugin and theme bugs published each day, sorted by what they mean for a small business. We refresh it twice a day. If you see something your website or office runs, tell us.
195 results · page 1 of 4
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115918 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. From 0.16.0 until 0.32.0, the /v1/responses endpoint allows an authenticated caller to place an attacker-controlled template in chat_template_kwargs[chat_template] when no server template is configured. ChatParams.get_apply_chat_template_kwargs() preserves that value without enforcing trust_request_chat_template, so the request can bypass the intended trust policy and cause CPU or memory exhaustion through Jinja rendering. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115917 · Other software · vllm-project · reported by GitHubLow 3.4
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowed_origins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM server on localhost, an internal network, or a VPN, call endpoints including GET /v1/models and POST /v1/chat/completions, consume GPU resources, and read responses. Cookie-based cross-site request forgery does not apply because credentials are disabled by default, and deployments requiring bearer authentication are exposed only when the caller already has a usable token. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115916 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, vllm/tokenizers/hf.py computes max_token_id with an off-by-one upper bound, allowing a pre-tokenized prompt ID equal to the vocabulary size to pass validation. For a tensor-parallel size of one and an exact multiple-of-64 vocabulary, VocabParallelEmbedding.forward does not mask that value before F.embedding uses it, causing a CUDA device-side assertion and terminating the engine. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115915 · Other software · vllm-project · reported by GitHubHigh 8.1
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, ImageMediaIO passes attacker-controlled image bytes to Pillow through Image.open(BytesIO(data)) without restricting decoder selection, and the MIME label does not constrain the selected format. Image-capable request paths through parse_image() and fetch_image() can therefore dispatch an EPS payload to host components. A host with Ghostscript affected by CVE-2024-29510 can execute commands with the vLLM process's privileges, while environments using Pillow 12.0.0 through 12.2.0 can trigger CVE-2026-59203 and exhaust media-loading workers. Deployments without the applicable vulnerable dependency are not affected by that corresponding impact. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115914 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. From 0.11.1 until 0.32.0, the DeepSeek-OCR, DeepSeek-OCR-2, and Unlimited-OCR image path allows a remote client that can reach the OpenAI-compatible API to submit a legal image whose normalized pixels sum to zero when that image is alone in its per-step encoder batch. _parse_and_validate_image_input returns no embedding, embed_multimodal propagates that value, and sanity_check_mm_encoder_outputs triggers a fatal assertion that shuts down the engine for all users. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115913 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the Phi-4 multimodal audio encoder in phi4mm_audio.py and phi4mm_utils.py uses eager attention by default, allowing a remote client to submit one legal long audio clip that causes MultiHeadedAttention.forward_attention to materialize and retain full attention matrices in self.attn. The scaled dot-product attention branch does not materialize these matrices. The resulting quadratic GPU-memory use can cause an out-of-memory condition and terminate EngineCore for all users. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115912 · Other software · vllm-project · reported by GitHubHigh 7.5
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, an authenticated caller can place forged shared-memory address and monotonic_id values in public multimodal features.kwargs_data. In scale-out deployments with shared-memory multimodal caching enabled, workers trust those values and pass attacker-selected bytes to pickle.loads(), allowing arbitrary Python execution even when insecure serialization is disabled. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-115911 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.32.0, authenticated callers can supply a request chat template when no server-side --chat-template is configured, or a malicious model repository can provide a template that is compiled and rendered by tokenizer.apply_chat_template() without a wall-clock timeout, output-size limit, or iteration bound. Rendering runs in a ThreadPoolExecutor with one worker by default, so a nonterminating render can block /v1/responses and other chat-template users after the request ends, while an output-expanding render can exhaust server memory. Harmony models are not affected because they do not use this Jinja path. This issue is fixed in version 0.32.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 25, 2026SiYuan 3.6.0 before 3.8.7 accepts and persists the networkServeTLS setting but never enforces it, so the kernel port keeps serving full API access over plain HTCVE-2026-116166 · Other software · siyuan-note · reported by VulnCheckHigh 7.6
SiYuan 3.6.0 before 3.8.7 accepts and persists the networkServeTLS setting but never enforces it, so the kernel port keeps serving full API access over plain HTTP alongside HTTPS. Network-adjacent attackers can sniff the cleartext login flow and session cookie, which lacks the Secure attribute while responses omit HSTS, to hijack kernel API sessions.
- Product
- siyuan
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Oct 20, 2026vLLM after 0.24.0 and before 0.31.0 contains a memory exhaustion vulnerability in MOSS-Audio model serving that caches a processor for every distinct request-coCVE-2026-114052 · Other software · vllm-project · reported by VulnCheckHigh 8.7
vLLM after 0.24.0 and before 0.31.0 contains a memory exhaustion vulnerability in MOSS-Audio model serving that caches a processor for every distinct request-controlled mm_processor_kwargs value in an unbounded dictionary. Unauthenticated attackers can send distinct or oversized mel_config values to the /tokenize endpoint, which bypasses --api-key, to exhaust host memory and crash the API process.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Oct 12, 2026Zebra before 6.4.0 contains an uncontrolled resource consumption vulnerability that allows unauthenticated peers to force full proof verification by serving forCVE-2026-109365 · Other software · ZcashFoundation · reported by VulnCheckMedium 6.3
Zebra before 6.4.0 contains an uncontrolled resource consumption vulnerability that allows unauthenticated peers to force full proof verification by serving forged block bodies with valid header hashes. Attackers can flip authorizing-only bytes such as a coinbase scriptSig, wasting Sapling, Orchard, and Ironwood proof verification without penalty and stalling sync for 67 seconds.
- Product
- zebra
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Oct 12, 2026Contrast 1.8.0 before 1.24.1 contains an improper verification vulnerability in Coordinator peer recovery that fails to confirm the serving peer uses the same mCVE-2026-109359 · Other software · edgelesssys · reported by VulnCheckHigh 8.7
Contrast 1.8.0 before 1.24.1 contains an improper verification vulnerability in Coordinator peer recovery that fails to confirm the serving peer uses the same manifest generation. Attackers with write access to manifest history can replay a superseded history head so withdrawn workloads obtain trusted mesh certificates, or with the secret seed disclose the Mesh CA key.
- Product
- contrast
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Oct 6, 2026An insecure direct object reference (IDOR) vulnerability in a CloudVision CUE file-serving interface may allow an authenticated network user, under specific attCVE-2026-102157 · Other software · Arista Networks · reported by psirt@arista.comMedium 6
An insecure direct object reference (IDOR) vulnerability in a CloudVision CUE file-serving interface may allow an authenticated network user, under specific attack conditions, to access another user's transient data.
- Product
- CloudVision CUE
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105760 · Other software · vllm-project · reported by GitHubMedium 5.3
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, a caller can use the request-level media_io_kwargs field to select the GLMGA video backend and supply large values for the fps and max_frames options without a strict work ceiling. GLMGA constructs and deduplicates an attacker-sized pre-decode frame-index list, allowing a compact request and tiny valid video to consume disproportionate CPU time and memory in the shared media-loading executor. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105759 · Other software · vllm-project · reported by GitHubMedium 5.9
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the Rust frontend's track_http_metrics middleware records the raw HTTP method token as a Prometheus label for requests reaching registered routes. An unauthenticated attacker can send unique arbitrary method tokens to unguarded routes such as /tokenize, causing Prometheus's Family::get_or_create function to permanently create counter and histogram label sets. Those label sets increase process memory usage and enlarge the /metrics response until the service or monitoring path is exhausted. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105758 · Other software · vllm-project · reported by GitHubMedium 5.3
vLLM is an inference and serving engine for large language models. From 0.24.0 until 0.30.0, the Qwen2VLVideoBackend and Qwen3VLVideoBackend classes accept request-level values for the media_io_kwargs.video.max_frames and media_io_kwargs.video.fps fields without enforcing server-side ceilings. An unauthenticated caller can submit these values to the /tokenize endpoint, causing the sampler to decode every frame selected from attacker-controlled video input, consume disproportionate frontend memory, and potentially terminate the API process before scheduling or admission control. The Rust frontend is not affected because it rejects the media_io_kwargs field. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105757 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, structured-output request failures can escape request-scoped validation and reach the EngineCore fatal-error path. A per-request backend mismatch can re-raise a grammar compilation exception, padding produced by the ngram_gpu speculative-decoding mode can pass a negative token to guidance validation, and the Rust frontend can admit empty structured-output values that the Python frontend rejects, allowing ordinary constrained-generation requests to terminate the shared engine. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105756 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, OpenAI-compatible request models accept a non-empty cache_salt value without enforcing the character and length restrictions required by the IPCCacheServerKey consumer in LMCache-MP. On deployments using the LMCache-MP connector, a salt that contains a forbidden character or exceeds the permitted length can raise an uncaught ValueError during scheduler cache lookup, causing EngineCore to terminate and denying service to all concurrent users. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105755 · Other software · vllm-project · reported by GitHubMedium 4.2
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, flash late-interaction scoring at the /score and /rerank endpoints derives each worker's query_key value from the caller-controlled X-Request-Id header. A concurrent request that reuses a victim's identifier can overwrite the cached query embedding so the victim's documents are scored against the attacker's query, and shared use counters can also cause a late-interaction cache-miss error. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105754 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the /inference/v1/generate endpoint in the disaggregated scale-out path accepts caller-supplied tensors in the features.kwargs_data field, cache identifiers in the features.mm_hashes field, ranges in the features.mm_placeholders field, and wire-selected multimodal field processors without rebinding them to the active model renderer contract. Forged grid geometry, field types, or non-positive placeholder lengths can terminate the shared EngineCore; when an attacker knows or can induce a victim's content hash, forged cache hashes can poison or retrieve cross-request encoder-cache state; and dropped sparse placeholder masks can alter replayed transport semantics. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105753 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.28.0, the default mirrored multimodal LRU cache can commit a media hash in the frontend sender cache during multimodal rendering and before engine admission, while the engine receiver cache never receives the payload if that request is rejected. A later request reusing the same media hash causes MultiModalProcessorSenderCache to send no payload and MultiModalReceiverCache to reach an assertion with the message "Expected a cached item," producing a shared-service availability failure. This issue is fixed in version 0.28.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 5, 2026vLLM is an inference and serving engine for large language models.CVE-2026-105752 · Other software · vllm-project · reported by GitHubLow 3.1
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, Harmony tool continuations submitted through "POST /v1/responses" requests rebuild the next-turn engine input without preserving the cache_salt value, placing the continuation prefix in the global unsalted cache namespace even when the caller enabled salting. On deployments with prefix caching enabled, which is the default, an authenticated tenant who can reconstruct a victim's low-entropy post-tool history can submit the same continuation and use the cached_tokens_per_turn count to determine whether the prefix was previously processed, defeating the intended tenant isolation of salted prefix caching. This issue is fixed in version 0.30.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Oct 2, 2026UTMStack before 11.2.16 contains an account enumeration vulnerability that allows unauthenticated attackers to determine registered email addresses by observingCVE-2026-82043 · Other software · UTMStack · reported by VulnCheckMedium 6.9
UTMStack before 11.2.16 contains an account enumeration vulnerability that allows unauthenticated attackers to determine registered email addresses by observing differing HTTP responses from the POST /api/account/reset-password/init endpoint. Attackers can submit arbitrary email addresses and distinguish registered accounts, which return 200 OK, from unregistered accounts, which trigger a 500 Internal Server Error with backend error details, enabling targeted phishing or credential attacks.
- Product
- UTMStack
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 28, 2026Payy is an Ethereum L2 zk-rollup for privacy preserving and regulatory compliant transactions.CVE-2026-48100 · Other software · polybase · reported by GitHubHigh 8.7
Payy is an Ethereum L2 zk-rollup for privacy preserving and regulatory compliant transactions. Prior to version 1.3.0, agg_agg forwards the compacted message stream from its inner proofs into a public messages: [Field; 1000] array, but it never checks that the unused tail of the outer array is zero. A registered prover can build a valid agg_final proof for an approved rollup block while inserting an extra burn message after the real messages. RollupV1.verifyRollup() then parses that public input as a normal burn and transfers USDC from the rollup contract to the attacker. This is a severe circuit soundness failure: the proof system accepts a public statement whose messages array is not fully derived from the verified inner proofs. On the current deployment, verifyRollup() is restricted to the existing allowlisted prover, so a fresh public caller cannot submit the invalid proof directly. That gate limits who can reach L1 today; it does not make the circuit statement sound. The issue becomes permissionless under the prover model described in the Payy whitepaper. Section 3.3.2 states: "To join as a prover, the prover is required to submit a small stake", and Section 3.3.1 states that
- Product
- payy
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 28, 2026mH-DEVELOPER smart home module contains a hardcoded SSH public key in /root/.ssh/authorized_keys, serving as a potential backdoor.CVE-2026-82928 · Other software · F&F Filipowski · reported by cvd@cert.plHigh 7.7
mH-DEVELOPER smart home module contains a hardcoded SSH public key in /root/.ssh/authorized_keys, serving as a potential backdoor. The SSH daemon allows root login via key authentication and starts automatically. An attacker with the matching private key can gain a root shell on any affected device, resulting in full system compromise. The key cannot be removed without remounting the file system and survives a factory reset. Vendor notes that this functionality was used only for service purposes. This issue was fixed in version 3.0.30
- Product
- mH-DEVELOPER
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 26, 2026vLLM is an inference and serving engine for large language models.CVE-2026-100653 · Other software · vllm-project · reported by VulnCheckHigh 8.3
vLLM is an inference and serving engine for large language models. In versions from 0.22.1 through 0.28.0, the operator-supplied model revision pin (--revision / --code-revision) is not propagated to several Hugging Face artifact loads for the FunAudioChat and Tarsier2 architectures: the WhisperFeatureExtractor and speech_tokenizer PreTrainedTokenizerFast loads in vllm/model_executor/models/funaudiochat.py and the Qwen2VLConfig.from_pretrained call used by Tarsier2ProcessingInfo in vllm/model_executor/models/qwen2_vl.py. As a result, deployments pinned to a reviewed revision still resolve these behavior-affecting processor, tokenizer, and config artifacts from the repository's default revision, so a later change to the upstream default branch can alter audio preprocessing, speech tokenizer behavior, or Tarsier2 configuration without any change to the operator's configured pin. This is a supply-chain integrity and reproducibility failure for pinned deployments; it is residual to the earlier fix tracked as GHSA-3ww4-5jv9-j5gm / CVE-2026-47155 and does not constitute remote code execution or a trust_remote_code=False bypass. The issue is fixed in version 0.28.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 26, 2026vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate.CVE-2026-100651 · Other software · vllm-project · reported by VulnCheckHigh 7.1
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Sep 26, 2026capgo.app through 12.129.0 fails to verify deletion status when serving cached bundle artifacts from the public file read endpoint.CVE-2026-100622 · Other software · Cap-go · reported by VulnCheckHigh 8.7
capgo.app through 12.129.0 fails to verify deletion status when serving cached bundle artifacts from the public file read endpoint. Unauthenticated attackers can download deleted bundles using cached URLs and trigger restoration of deleted objects into R2 storage on cache hits.
- Product
- capgo.app
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Sep 25, 2026@openclaw/whatsapp (npm) before 2026.8.1 exposes the WhatsApp login tool through the generic channel-tool path without preserving the originating sender's ownerCVE-2026-100532 · Other software · openclaw · reported by VulnCheckHigh 7.2
@openclaw/whatsapp (npm) before 2026.8.1 exposes the WhatsApp login tool through the generic channel-tool path without preserving the originating sender's owner status, so the owner-only tool boundary is not enforced. An admitted non-owner sender able to steer the tool can request a forced login and receive a new QR code for a configured account, disconnecting the Gateway's WhatsApp account and causing loss of availability; full account relinking additionally requires the attacker to scan the returned QR code with another phone. The issue affects the owner-only tool boundary rather than WhatsApp transport authentication. Fixed in 2026.8.1.
- Product
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Sep 18, 2026LMDeploy is a toolkit for compressing, deploying, and serving large language models.CVE-2026-33625 · Other software · InternLM · reported by GitHubHigh 8.8
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions 012.1 through 0.12.2 contain a code injection vulnerability in `lmdeploy/pytorch/config.py` line 620 that allows an attacker to execute arbitrary Python code by publishing a malicious HuggingFace model with a crafted `quantization_config.quant_dtype` value. When a user loads the model with lmdeploy, the `quant_dtype` is passed to `eval(f'torch.{quant_dtype}')` without any validation. Version 0.12.3 contains a patch.
- Product
- lmdeploy
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 18, 2026LMDeploy is a toolkit for compressing, deploying, and serving large language models.CVE-2025-66455 · Other software · InternLM · reported by GitHubCritical 9.8
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received through a ZeroMQ PULL socket. PyZMQ implements `recv_pyobj()` using Python pickle deserialization, which can execute arbitrary code while reconstructing an object. The peer address used by the receiver was supplied through the `POST /distserve/p2p_connect` HTTP endpoint. An attacker who could reach an affected DistServe API server could cause the server to connect to an attacker-controlled ZeroMQ endpoint and deserialize a crafted pickle payload. API-key authentication is not enabled unless the operator explicitly configures it. As a result, affected DistServe deployments without API keys allowed unauthenticated remote code execution with the privileges of the LMDeploy serving process. This issue affects the PyTorch backend when PD-disaggregation/DistServe is enabled. Ordinary deployments that do not use the affected disaggregated-serving path do not expose this data flow. The fix was released in LMDeploy 0.16.0. Users who cannot up
- Product
- lmdeploy
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 17, 2026In the Linux kernel, the following vulnerability has been resolved: ntfs: validate usa_ofs before preserving the update sequence number When ntfs_mft_record_aCVE-2026-90117 · Computers, phones & browsers · Linux · reported by 416baaa9-dc9f-4396-8d5f-8c081fb06d67
In the Linux kernel, the following vulnerability has been resolved: ntfs: validate usa_ofs before preserving the update sequence number When ntfs_mft_record_alloc() reuses a free mft record it reads the old update sequence number straight from the on-disk record: usn = *(__le16 *)((u8 *)m + le16_to_cpu(m->usa_ofs)); Here m points into the raw $MFT page-cache folio, which still holds unvalidated, MST-protected bytes: the folio is read by a plain iomap_read_folio() and neither post_read_mst_fixup() nor ntfs_mft_record_check() has run on it (both work on private copies). m->usa_ofs is therefore an untrusted u16, and a corrupted record can put it past the end of the record so the two-byte read lands outside the folio. Reading such a record while creating a file gives, under KASAN: BUG: KASAN: use-after-free in ntfs_mft_record_alloc+... Read of size 2 at addr ... ntfs_mft_record_alloc -> __ntfs_create -> ntfs_create -> path_openat Only preserve the old update sequence number when usa_ofs is even and in range, mirroring the check ntfs_mft_record_check() already applies; otherwise leave usn zero, which the existing restore below skips.
- Product
- Linux
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgit.kernel.orggit.kernel.org
Sep 16, 2026vLLM is an inference and serving engine for large language models.CVE-2026-69147 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set media_io_kwargs.video.video_backend to pynvvideocodec, and MediaConnector.fetch_video forwards that choice to VideoMediaIO even when startup configuration selected a software decoder. The engine's _reserve_mm_ipc_gpu_memory logic budgets decoder memory only from static configuration, so the request-selected VIDEO_LOADER_REGISTRY backend can create a CUDA context, decoder surfaces, and decoded-frame allocations that were not removed from the engine's KV-cache budget. An attacker able to submit video requests to a video-capable GPU deployment with PyNvVideoCodec installed can exhaust shared GPU memory, causing request failures, worker crashes, or denial of service. The first release containing the fix is version 0.28.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 16, 2026vLLM is an inference and serving engine for large language models.CVE-2026-57173 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the input_audio handling path for /v1/chat/completions calls AudioMediaIO.load_bytes or AudioMediaIO.load_file without passing VLLM_MAX_AUDIO_DECODE_DURATION_S to the shared audio decoder. An unauthenticated client can therefore submit a small compressed audio input that expands into a very large float32 PCM allocation, bypassing the duration guard already used by /v1/audio/transcriptions and causing an out-of-memory worker crash. Inline data URLs reach this path without being bounded by VLLM_AUDIO_FETCH_TIMEOUT. The issue affects deployments serving an audio-capable model, and authentication changes only the deployment-specific reachability. This issue is fixed in version 0.24.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 16, 2026LMDeploy is a toolkit for compressing, deploying, and serving large language models.CVE-2025-59953 · Other software · InternLM · reported by GitHubCritical 9.8
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.1 and prior to version 0.10.2, the LMdeploy implements an rpc server (AsyncRPCServer in zmq_rpc.py) for supporting the RPC communications. In its core functionality call_and_response(), I found it will directly use the pickles.loads() to deserialize the received messages without any sanitization, hence resulting in a remote code execution vulnerability by this RPC server. Version 0.10.2 contains a patch.
- Product
- lmdeploy
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 7, 2026Affected versions of MISP do not enforce parent-event visibility when serving cryptographic keys through CryptographicKeysController::view().CVE-2026-86408 · Other software · MISP · reported by 5a6e4751-2f3f-4070-9419-94fb35b644e8High 7.1
Affected versions of MISP do not enforce parent-event visibility when serving cryptographic keys through CryptographicKeysController::view(). The vulnerable handler queried CryptographicKey directly using the supplied key ID and selected sensitive fields such as: * type * key_data * fingerprint but did not fetch or authorize the associated parent event first. The upstream commit explicitly states that cryptographicKeys/view could return a protected event’s signing key to any authenticated user. The fix adds parent_id and parent_type to the lookup and then enforces authorization through the associated event using fetchSimpleEvent($user, parent_id). If the parent is not an Event, access is limited to site administrators. Version affected: ≤2.5.45
- Product
- MISP
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 4, 2026Documenso 2.17.0 contains an access control vulnerability in the PDF-serving endpoint that fails to validate document visibility settings.CVE-2026-85697 · Other software · documenso · reported by VulnCheckHigh 7.1
Documenso 2.17.0 contains an access control vulnerability in the PDF-serving endpoint that fails to validate document visibility settings. Attackers with low privileges can read restricted documents within their team or cross-tenant by leveraging missing ownership validation on document data identifiers.
- Product
- documenso
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Sep 2, 2026SiYuan before v3.8.2 contains a stored cross-site scripting vulnerability in asset serving due to an incomplete extension blocklist that misses script-capable fCVE-2026-84803 · Other software · siyuan-note · reported by VulnCheckHigh 8.6
SiYuan before v3.8.2 contains a stored cross-site scripting vulnerability in asset serving due to an incomplete extension blocklist that misses script-capable file types. Attackers can upload files with extensions like .xht, .ehtml, .xsl, .xbl, or .rdf that resolve to executable media types and execute JavaScript to steal API tokens and compromise workspaces.
- Product
- siyuan
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Sep 1, 2026A flaw was found in pulpcore's content serving application.CVE-2026-84232 · Other software · Red Hat · reported by Red HatMedium 5.4
A flaw was found in pulpcore's content serving application. Files uploaded to Pulp file-type repositories are served with their original content type (e.g., text/html for .html files, image/svg+xml for .svg files) and without a Content-Disposition: attachment header when using local filesystem storage. An authenticated user or attacker with content upload permissions can upload a specially crafted HTML or SVG file containing JavaScript, which executes in the browser of any user who visits the file URL, resulting in stored cross-site scripting (XSS) in the context of the host application.
- Product
- Red Hat Ansible Automation Platform 2
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordaccess.redhat.combugzilla.redhat.com
Aug 27, 2026JetBackup: Does not perform its multisite authorisation check before serving backup archives and job logsCVE-2026-19454 · WordPress plugins & themes · reported by WPScanMedium 4.4
The JetBackup WordPress plugin before 3.1.23.5 does not perform its multisite authorisation check before serving backup archives and job logs, allowing an administrator of the network's main site who is not a Super Admin to download a full backup of the entire network, including every site's data and the shared webroot.
- Product
- JetBackup
- Fixed in
- 3.1.23.5
- What to do
- Update JetBackup to 3.1.23.5 or newer.
Aug 22, 2026In the Linux kernel, the following vulnerability has been resolved: rqspinlock: Reset tail when preserving queue on deadlock Currently, the destruction of theCVE-2026-74686 · Computers, phones & browsers · Linux · reported by 416baaa9-dc9f-4396-8d5f-8c081fb06d67
In the Linux kernel, the following vulnerability has been resolved: rqspinlock: Reset tail when preserving queue on deadlock Currently, the destruction of the waiter queue is suppressed for rqspinlock in cases where a deadlock is detected. Deadlock checks happen relatively frequently (on entry for AA, within 1ms for ABBA), and waiter threads may not be involved in locking scenarios involving deadlocks. Thus, it is useful to not flush the queue and let other waiters take a stab at acquiring the lock after we detect a deadlock and exit. However, we need to follow the same logic as what we did previously for the waitq_timeout label: reset the tail, and if we cannot, signal the next waiter appropriately. In case of deadlocks, this signal would just mark the MCS node as unlocked, and in case of timeouts, it would signal RES_TIMEOUT_VAL. The difference thus is in the value propagated, which decides whether the queue remains active or gets flushed. Not doing the tail reset, and waiting for the next waiter can lead to cases where we are the final waiter, and thus no next waiter arrives, leading to intermittent stalls in this path. Once the next waiter does join, we will be unblocked. I
- Product
- Linux
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgit.kernel.orggit.kernel.org
Aug 19, 2026LMDeploy deserializes disaggregated-serving peer messages with pickle.CVE-2026-76850 · Other software · InternLM · reported by VulnCheckCritical 9.3
LMDeploy deserializes disaggregated-serving peer messages with pickle. The handle_zmq_recv coroutine in lmdeploy/pytorch/disagg/conn/engine_conn.py reads peer-to-peer cache-free requests with recv_pyobj(), which deserializes the received bytes with pickle.loads(), and the isinstance check against DistServeCacheFreeRequest runs only after deserialization has already completed. The peer that supplies those bytes is caller-controlled: p2p_connect passes remote_engine_endpoint_info.zmq_address from the request body to connect() on the ZMQ PULL socket, and the POST /distserve/p2p_initialize and /distserve/p2p_connect endpoints in lmdeploy/serve/openai/api_server.py apply no authentication unless the server is started with api_keys, which defaults to None. A remote attacker can direct an engine to pull from a ZMQ endpoint under their control and execute arbitrary code in the engine process. Deployments that do not enable disaggregated serving are not affected, because the receive loop is only started once the migration backend accepts the connection.
- Product
- lmdeploy
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 17, 2026vLLM is an inference and serving engine for large language models.CVE-2026-73560 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the MiMoV2OmniMultiModalProcessor in vllm/transformers_utils/processors/mimo_v2_omni.py passes attacker-controlled image and audio strings through _fetch_image, requests.get, and Image.open instead of MediaConnector, bypassing allowed_media_domains and allowed_local_media_path protections and allowing server-side requests and reads of arbitrary files accessible to the vLLM process. This issue is fixed in version 0.26.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 17, 2026vLLM is an inference and serving engine for large language models.CVE-2026-71486 · Other software · vllm-project · reported by GitHubMedium 4.3
vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects whose generate_responses, choices, token_ids, prompt_logprobs, logprobs.content, top_logprobs, and routed_experts structures are processed by OnlineDerenderer and tokenizer.decode before max_model_len, max_tokens, max_num_seqs, or response-size limits are enforced, allowing an authenticated API client to consume excessive CPU and memory and produce oversized responses. This issue is fixed in version 0.26.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 17, 2026SiYuan before v3.7.4 fails to set Content-Disposition and X-Content-Type-Options headers when serving arbitrary file assets, allowing stored cross-site scriptinCVE-2026-74800 · Other software · siyuan-note · reported by VulnCheckCritical 9.4
SiYuan before v3.7.4 fails to set Content-Disposition and X-Content-Type-Options headers when serving arbitrary file assets, allowing stored cross-site scripting attacks. Authenticated attackers can upload HTML files as assets and execute scripts with full kernel API access when the workspace owner opens the asset link.
- Product
- siyuan
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Full record and what to do →NVD recordgithub.comvulncheck.com
Aug 16, 2026Extra Product Options Builder for WooCommerce: Does not verify that the requester is entitled to a customer-uploaded file before serving itCVE-2026-19728 · WordPress plugins & themes · reported by WPScanHigh 7.5
The Extra Product Options Builder for WooCommerce WordPress plugin before 1.2.176 does not verify that the requester is entitled to a customer-uploaded file before serving it, allowing unauthenticated users who obtain a file's stored name to retrieve it. The Extra Product Options Builder for WooCommerce WordPress plugin before 1.2.176 writes a deny-all rule into its upload directories, so the disclosure only crosses a boundary on web servers that honour it, such as Apache. Where it is ignored, as on a default nginx setup, the same files are already served at their direct URL and the endpoint exposes nothing further.
- Product
- Extra Product Options Builder for WooCommerce
- Fixed in
- 1.2.176
- What to do
- Update Extra Product Options Builder for WooCommerce to 1.2.176 or newer.
Aug 13, 2026A malicious GOSUMDB was capable of serving arbitrary module content not contained within the transparency log.CVE-2026-56864 · Other software · Go toolchain · reported by security@golang.orgHigh 7.5
A malicious GOSUMDB was capable of serving arbitrary module content not contained within the transparency log. This attack allows for a coordinating GOPROXY and GOSUMDB to serve a client malicious module content that cannot be detected by evaluating the transparency log. In order to determine if you have been affected: rm -r go.sum go.work.sum vendor/ && go mod tidy
- Product
- cmd/go
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 13, 2026vantage6 is an open-source infrastructure for privacy preserving analysis.CVE-2026-73652 · Other software · vantage6 · reported by GitHubHigh 7.1
vantage6 is an open-source infrastructure for privacy preserving analysis. In version 5.0.2 and earlier, the algorithm-store edit permission lacks an ownership check, allowing one algorithm developer to alter another developer's algorithm while it is pending or under review. The attacker can change metadata including the algorithm image or image tag, causing reviewers and nodes to trust a different image from the one originally submitted for approval. No fixed version is available as of this review.
- Product
- vantage6
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 13, 2026vLLM is an inference and serving engine for large language models.CVE-2026-73559 · Other software · vllm-project · reported by GitHubMedium 6.5
vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.26.0, the /v1/completions CompletionRequest.prompt field in vllm/entrypoints/openai/completion/protocol.py accepts an unbounded list[str] or list[list[int]], prompt_to_seq() in vllm/renderers/inputs/preprocess.py and OnlineRenderer.preprocess_completion() in vllm/renderers/online_renderer.py expand every element, and vllm/entrypoints/openai/completion/serving.py creates one engine generator and response slot per prompt, allowing an authenticated API client to exhaust CPU, memory, async scheduling capacity, engine request slots, and response buffering with one request. This issue is fixed in version 0.26.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Aug 13, 2026vLLM is an inference and serving engine for large language models.CVE-2026-73558 · Other software · vllm-project · reported by GitHubMedium 5.3
vLLM is an inference and serving engine for large language models. Prior to 0.27.0, an integer overflow in blockIdx.x * 2 * d in activation_kernels.cu can cause act_and_mul_kernel to consume another batched user's input, allowing a request processed in the same inference batch to receive a partial or complete copy of another user's inference result. This issue is fixed in version 0.27.0.
- Product
- vllm
- What to do
- Apply the vendor’s update. The NVD record links to the advisory.
Where this comes from: the exploited list is CISA’s Known Exploited Vulnerabilities catalog. Everything else is the whole National Vulnerability Database, where each record is published by the team that found the bug (Wordfence, Patchstack, WPScan, the vendors themselves and others); we show who. Exploit odds are FIRST’s EPSS, updated daily. Titles and categories are ours, so check the linked record before acting on a detail. For a write-up of the big ones, see our advisories. Every vendor, and every WordPress plugin and theme with a recorded flaw, also has its own page. All of it is free to reuse as JSON, CSV or RSS.
Weekly security brief
The flaws that hit local websites, every Friday.
One short email: what got exploited this week, which WordPress plugins broke, and what to check. Unsubscribe in one click.
Tell us what you're building.
A website, your marketing, a security check, an app, or the thing you've been running out of a spreadsheet for three years. A few lines is plenty.
Not ready to talk yet? Run the free website check and see what Google sees.