Briefs

CVE-2026-44477 allows a low-privileged database user to escalate to PostgreSQL superuser and achieve OS command execution as the `postgres` user within the primary pod by exploiting the metrics exporter's superuser connection via custom metric queries or the default configuration.

Next.js applications using Partial Prerendering through Cache Components are vulnerable to connection exhaustion (CVE-2026-44579), where crafted POST requests to a server action trigger a request-body handling deadlock, consuming server capacity and leading to denial of service.

Next.js applications using the Pages Router with `i18n` and middleware-based authorization are vulnerable to an authentication bypass (CVE-2026-44573), allowing unauthorized access to protected page data via locale-less `/_next/data/<buildId>/<page>.json` requests.

A vulnerability in Next.js (CVE-2026-44574) allows for authorization bypass in applications that use middleware to protect dynamic routes, enabling attackers to render protected content without proper authorization by crafting specific query parameters.

Next.js applications using WebSocket upgrades are vulnerable to server-side request forgery (SSRF) through crafted WebSocket upgrade requests, allowing attackers to proxy requests to internal or external destinations, affecting self-hosted applications running versions npm/next (>= 13.4.13, < 15.5.16) and npm/next (>= 16.0.0, < 16.2.5).

Ubuntu released security notices between May 4 and 10, 2026, addressing vulnerabilities in the Linux kernel affecting Ubuntu 20.04 LTS, 22.04 LTS, 24.04 LTS, and 25.10, requiring timely updates.

Ella Core is vulnerable to UE downlink redirection (CVE-2026-44473) due to missing SCTP association verification, enabling a malicious radio to forge a PDUSessionResourceSetupResponse and redirect downlink traffic.

The FileSystemLoader in python-liquid versions before 2.2.0 allows malicious template authors to read arbitrary files outside the search paths via the `{% include %}` and `{% render %}` tags by using absolute paths; this is resolved in version 2.2.0 by checking for absolute paths in the `resolve_path()` method.

Sensitive headers (`Authorization`, `Cookie`, and `Proxy-Authorization`) are forwarded across origins in proxied low-level redirects when using `HTTPConnection.urlopen()` instances created via `ProxyManager.connection_from_url()` in urllib3 versions before 2.7.0, potentially exposing credentials to unintended third parties; upgrade to version 2.7.0 or later to remediate this issue.

Urllib3 versions before 2.7.0 are vulnerable to excessive resource consumption when using the streaming API to decompress responses, particularly when using the Brotli library or calling HTTPResponse.drain_conn() after partial decompression, leading to high CPU usage and memory allocation, potentially causing a denial-of-service condition (CVE-2026-44432).

@vitejs/plugin-rsc is vulnerable to a denial-of-service attack due to an embedded vulnerable version of react-server-dom-webpack, potentially causing resource exhaustion.

go-git may parse malformed Git objects differently than upstream Git, leading to inconsistent interpretation and potentially allowing the signing or verification of commits with altered metadata, as described in CVE-2026-45022.

GuardDog versions 1.0.0 through 2.9.0 are vulnerable to Server-Side Request Forgery (SSRF) and potential `GH_TOKEN` exfiltration due to a blind URL rewrite in remote project scanning; an attacker can influence the scanned repository URL to trigger SSRF and capture the `GH_TOKEN` used by GuardDog.

A malformed HTTP request can crash any Node.js process running the OpenTelemetry JS Prometheus exporter. The metrics endpoint has no error handling around URL parsing, so a request with an invalid URI causes an uncaught `TypeError` that terminates the process, leading to a denial of service. Update `@opentelemetry/exporter-prometheus` and `@opentelemetry/sdk-node` to version **0.217.0** or later and `@opentelemetry/auto-instrumentations-node` to version **0.75.0** or later to remediate.

Open WebUI versions before 0.6.19 have inconsistent authorization controls within the memories API, allowing standard users to view, delete, and restore other users' memories, potentially leading to sensitive data disclosure and unauthorized access as tracked by CVE-2026-44570.

Dozzle is vulnerable to Cross-Site WebSocket Hijacking (CSWSH) due to a permissive CheckOrigin configuration and the use of SameSite=Lax for JWT cookies, allowing attackers on the same site to gain shell access to containers even with authentication enabled, tracked as CVE-2026-44985.

Open WebUI version v0.3.10 has a CORS misconfiguration and session validation issue that can lead to remote code execution due to a one-click attack against admin users.

Open WebUI is vulnerable to path traversal (CVE-2026-44565), allowing attackers to upload files to arbitrary locations on the web server's filesystem and subsequently delete them due to insufficient filename sanitization in the `/ollama/models/upload` API endpoint.

PraisonAI resolves tool names against module globals and `__main__` after failing to match declared tools, allowing an attacker who can influence tool-call names to invoke unintended application callables, leading to potential unauthorized state changes and command execution.

PraisonAI versions 2.7.2 through 4.6.35 are vulnerable to an arbitrary file write due to improper validation of symlinks during archive extraction, affecting `recipe pull`, `recipe publish`, and `recipe unpack` flows.

PraisonAI's MCP server is vulnerable to path traversal leading to arbitrary code execution by writing a Python `.pth` file into the user's site-packages directory, triggered via poisoned LLM contexts or unauthenticated HTTP-stream transports due to unvalidated kwargs in the dispatcher and lack of containment checks in file-handling tools.

PraisonAI ships a legacy Flask API server with authentication disabled by default, allowing any reachable caller to access `/agents` and trigger the configured `agents.yaml` workflow through `/chat` without providing a token (CVE-2026-44338).

Threat actors are leveraging AI to enhance vulnerability discovery, exploit development, defense evasion, and autonomous operations, with state-sponsored groups showing particular interest in AI-driven vulnerability research and exploit generation.

Multiple vulnerabilities in Spring products could allow a remote attacker to execute arbitrary code, cause a denial of service, or breach data confidentiality.

A remote, authenticated attacker can exploit multiple vulnerabilities in Apache Airflow Providers OpenSearch and Elasticsearch to disclose sensitive information.

Multiple vulnerabilities in KDE Kdenlive and Okular allow a remote, anonymous attacker to execute arbitrary code, bypass security measures, manipulate data, disclose confidential information, or cause a denial-of-service condition.

A local attacker can exploit a vulnerability in jq to bypass security measures.

A remote, anonymous attacker can exploit a vulnerability in vm2 to execute arbitrary code, potentially leading to arbitrary code execution on the host system.

A remote, authenticated attacker can exploit multiple vulnerabilities in HCL BigFix WebUI applications to disclose sensitive information.

A remote, authenticated attacker can exploit a vulnerability in Rancher Fleet Helm Deployer to bypass security measures and disclose sensitive information, which may enable further attacks.