OpenAirInterface v2.2.0 accepts Security Mode Complete without any integrity protection. Configuration has supported integrity NIA1 and NIA2. But if an UE sends initial registration request with only security capability IA0, OpenAirInterface accepts and proceeds. This downgrade security context can lead to the possibility of replay attack.

openairinterface

OpenAirInterface Version 2.2.0 has a Buffer Overflow vulnerability in processing UplinkNASTransport containing Authentication Response containing a NAS PDU with oversize response (For example 100 byte). The response is decoded by AMF and passed to the AUSF component for verification. AUSF crashes on receiving this oversize response. This can prohibit users from further registration and verification and can cause Denial of Services (DoS).

openairinterface

In OpenAirInterface V2.2.0 AMF, Out of sequence messages causes incorrect state transition during UE registration procedure. This allows authentication to be bypassed completely. If a SecurityModeComplete message is sent after InitialUERegistration, a registration reject is received followed by a registration accept! This leads the UE to be registered without proper authentication.

openairinterface

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. Prior to 1.83.0, when JWT authentication is enabled (enable_jwt_auth: true), the OIDC userinfo cache uses token[:20] as the cache key. JWT headers produced by the same signing algorithm generate identical first 20 characters. This configuration option is not enabled by default. Most instances are not affected. An unauthenticated attacker can craft a token whose first 20 characters match a legitimate user's cached token. On cache hit, the attacker inherits the legitimate user's identity and permissions. This affects deployments with JWT/OIDC authentication enabled. Fixed in v1.83.0.

litellm

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. Prior to 1.83.0, the /config/update endpoint does not enforce admin role authorization. A user who is already authenticated into the platform can then use this endpoint to modify proxy configuration and environment variables, register custom pass-through endpoint handlers pointing to attacker-controlled Python code, achieving remote code execution, read arbitrary server files by setting UI_LOGO_PATH and fetching via /get_image, and take over other privileged accounts by overwriting UI_USERNAME and UI_PASSWORD environment variables. Fixed in v1.83.0.

litellm

vLLM is an inference and serving engine for large language models (LLMs). From 0.1.0 to before 0.19.0, a Denial of Service vulnerability exists in the vLLM OpenAI-compatible API server. Due to the lack of an upper bound validation on the n parameter in the ChatCompletionRequest and CompletionRequest Pydantic models, an unauthenticated attacker can send a single HTTP request with an astronomically large n value. This completely blocks the Python asyncio event loop and causes immediate Out-Of-Memory crashes by allocating millions of request object copies in the heap before the request even reaches the scheduling queue. This vulnerability is fixed in 0.19.0.

vllm

OpenAirInterface V2.2.0 AMF crashes when it receives an NGAP message with invalid procedure code or invalid PDU-type. For example when the message specification requires InitiatingMessage but sent with successfulOutcome.

openairinterface

Prompt injection vulnerability in 1millionbot Millie chatbot that occurs when a user manages to evade chat restrictions using Boolean prompt injection techniques (formulating a question in such a way that, upon receiving an affirmative response ('true'), the model executes the injected instruction), causing it to return prohibited information and information outside its intended context. Successful exploitation of this vulnerability could allow a malicious remote attacker to abuse the service for purposes other than those originally intended, or even execute out-of-context tasks using 1millionbot's resources and/or OpenAI's API key. This allows the attacker to evade the containment mechanisms implemented during LLM model training and obtain responses or chat behaviors that were originally restricted.

1millionbot

OpenAirInterface V2.2.0 AMF crashes when it fails to decode the message. Not all decode failures result in a crash. But the crash is consistent for particular inputs. An example input in hex stream is 80 00 00 0E 00 00 01 00 0F 80 02 02 40 00 58 00 01 88.

openairinterface

n8n is an open source workflow automation platform. Prior to versions 2.14.1, 2.13.3, and 1.123.27, an authenticated user with the `global:member` role could exploit chained authorization flaws in n8n's credential pipeline to steal plaintext secrets from generic HTTP credentials (`httpBasicAuth`, `httpHeaderAuth`, `httpQueryAuth`) belonging to other users on the same instance. The attack abuses a name-based credential resolution path that does not enforce ownership or project scope, combined with a bypass in the credentials permission checker that causes generic HTTP credential types to be skipped during pre-execution validation. Together, these flaws allow a member-role user to resolve another user's credential ID and execute a workflow that decrypts and uses that credential without authorization. Native integration credential types (e.g. `slackApi`, `openAiApi`, `postgres`) are not affected by this issue. This vulnerability affects Community Edition only. Enterprise Edition has additional permission gates on workflow creation and execution that independently block this attack chain. The issue has been fixed in n8n versions 1.123.27, 2.13.3, and 2.14.1. Users should upgrade to one of these versions or later to remediate the vulnerability. If upgrading is not immediately possible, administrators should consider the following temporary mitigations: Restrict instance access to fully trusted users only, and/or audit credentials stored on the instance and rotate any generic HTTP credentials (`httpBasicAuth`, `httpHeaderAuth`, `httpQueryAuth`) that may have been exposed. These workarounds do not fully remediate the risk and should only be used as short-term mitigation measures.

n8n

New API is a large language mode (LLM) gateway and artificial intelligence (AI) asset management system. Prior to version 0.11.4-alpha.2, an Insecure Direct Object Reference (IDOR) vulnerability in the video proxy endpoint (`GET /v1/videos/:task_id/content`) allows any authenticated user to access video content belonging to other users and causes the server to authenticate to upstream AI providers (Google Gemini, OpenAI) using credentials derived from tasks they do not own. The missing authorization check is a single function call — `model.GetByOnlyTaskId(taskID)` queries by `task_id` alone with no `user_id` filter, while every other task-lookup in the codebase enforces ownership via `model.GetByTaskId(userId, taskID)`. Version 0.11.4-alpha.2 contains a patch.

newapi

GROWI OpenAI thread/message API endpoints do not perform authorization. Affected are v7.4.5 and earlier versions. A logged-in user who knows a shared AI assistant's identifier may view and/or tamper the other user's threads/messages.

The Greenshift – animation and page builder blocks plugin for WordPress is vulnerable to Sensitive Information Exposure in all versions up to, and including, 12.8.3 via the automated Settings Backup stored in a publicly accessible file. This makes it possible for unauthenticated attackers to extract sensitive data including the configured OpenAI, Claude, Google Maps, Gemini, DeepSeek, and Cloudflare Turnstile API keys.

LangChain is a framework for building agents and LLM-powered applications. Prior to 1.2.11, the ChatOpenAI.get_num_tokens_from_messages() method fetches arbitrary image_url values without validation when computing token counts for vision-enabled models. This allows attackers to trigger Server-Side Request Forgery (SSRF) attacks by providing malicious image URLs in user input. This vulnerability is fixed in 1.2.11.

langchain

Typebot is an open-source chatbot builder. In versions prior to 3.13.2, client-side script execution in Typebot allows stealing all stored credentials from any user. When a victim previews a malicious typebot by clicking "Run", JavaScript executes in their browser and exfiltrates their OpenAI keys, Google Sheets tokens, and SMTP passwords. The `/api/trpc/credentials.getCredentials` endpoint returns plaintext API keys without verifying credential ownership. Version 3.13.2 fixes the issue.

typebot

The BetterDocs plugin for WordPress is vulnerable to Sensitive Information Exposure in all versions up to, and including, 4.3.3 via the scripts() function. This makes it possible for authenticated attackers, with contributor-level access and above, to extract sensitive data including the OpenAI API key stored in plugin settings.

OpenAirInterface CN5G AMF<=v2.0.1 There is a logical error when processing JSON format requests. Unauthorized remote attackers can send malicious JSON data to AMF's SBI interface to launch a denial-of-service attack.

openairinterface

OpenAirInterface CN5G AMF<=v2.1.9 has a buffer overflow vulnerability in processing NAS messages. Unauthorized remote attackers can launch a denial-of-service attack and potentially execute malicious code by accessing port N1 and sending an imsi string longer than 1000 to AMF.

openairinterface

The Tag, Category, and Taxonomy Manager – AI Autotagger with OpenAI plugin for WordPress is vulnerable to unauthorized modification of data due to a missing capability check on the taxopress_ai_add_post_term function in all versions up to, and including, 3.41.0. This makes it possible for authenticated attackers, with Contributor-level access and above, to add or remove taxonomy terms (tags, categories) on any post, including ones they do not own.

An authentication bypass vulnerability exists in AnythingLLM v1.8.5 in via the /api/workspaces endpoint. The endpoint fails to implement proper authentication checks, allowing unauthenticated remote attackers to enumerate and retrieve detailed information about all configured workspaces. Exposed data includes: workspace identifiers (id, name, slug), AI model configurations (chatProvider, chatModel, agentProvider), system prompts (openAiPrompt), operational parameters (temperature, history length, similarity thresholds), vector search settings, chat modes, and timestamps.

mintplexlabs

The Tag, Category, and Taxonomy Manager – AI Autotagger with OpenAI plugin for WordPress is vulnerable to time-based blind SQL Injection via the 'existing_terms_orderby' parameter in the AI preview AJAX endpoint in all versions up to, and including, 3.40.1. This is due to insufficient escaping on user-supplied parameters and lack of SQL query parameterization. This makes it possible for authenticated attackers, with Contributor-level access and above who have AI metabox permissions, to append additional SQL queries into already existing queries that can be used to extract sensitive information from the database, cause performance degradation, or enable data inference through time-based techniques.

The Tag, Category, and Taxonomy Manager – AI Autotagger with OpenAI plugin for WordPress is vulnerable to time-based SQL Injection via the "getTermsForAjax" function in all versions up to, and including, 3.40.1. This is due to insufficient escaping on the user supplied parameters and lack of sufficient preparation on the existing SQL query. This makes it possible for authenticated attackers, with contributor level access and above, to append additional SQL queries into already existing queries that can be used to extract sensitive information from the database granted they have metabox access for the taxonomy (enabled by default for contributors).

taxopress