The Gutenberg Blocks with AI by Kadence WP – Page Builder Features plugin for WordPress is vulnerable to unauthorized post publication in all versions up to, and including, 3.5.32 due to a misconfigured capability check on the 'get_items_permission_check' function permission callback of the 'process_pattern' REST API endpoint. This makes it possible for authenticated attackers, with Contributor-level access and above, to create and immediately publish posts of any type (including pages), bypassing the standard WordPress review workflow where contributors must submit posts for administrator approval.

Improper Neutralization of Parameter/Argument Delimiters vulnerability in elixir-plug plug allows an attacker to inject or override HTTP cookie attributes. The Plug.Conn.Cookies.encode/2 function in lib/plug/conn/cookies.ex builds the Set-Cookie response header by interpolating the cookie value and its path, domain, same_site, and extra attributes directly into the header without neutralizing the ';' delimiter that separates cookie attributes. An application that places attacker-controlled data into a cookie value or attribute (for example via Plug.Conn.put_resp_cookie/4 when reflecting a username or preference) lets an attacker inject a ';' to append or override cookie attributes (such as Domain and Path scope, or dropping the Secure and HttpOnly flags), enabling cookie tossing and session fixation. Carriage return, line feed, and null bytes are rejected by Plug.Conn header validation, so HTTP response splitting is not possible, but attribute injection through ';' is not prevented. This issue affects plug: from 0.1.0 before 1.16.6, from 1.17.0 before 1.17.4, from 1.18.0 before 1.18.5, from 1.19.0 before 1.19.5, from 1.20.0 before 1.20.3.

The charging station websocket endpoint accepts connections without proper authentication, which could lead to privilege escalation.

Grist is spreadsheet software using Python as its formula language. Prior to 1.7.15, several server-rendered Grist pages embedded user-controlled values into the page and into inline scripts without fully escaping them, allowing cross-site scripting. On the main application page, a document's name or description, set by a document editor, is rendered into the page that other users load when opening the document. On the OAuth2 end-of-flow page, the openerOrigin request parameter was reflected back into the served page. Injected script runs in the victim's Grist origin and can act through the authenticated session, reading or modifying data and changing sharing settings and access rules. A document editor could therefore escalate to owner-level access. This issue is fixed in version 1.7.15.

PraisonAI (praisonaiagents) before 1.6.78 contains a path traversal vulnerability in the FastContext feature (praisonaiagents.context.fast). FastContextAgent.execute_tool() prepends the configured workspace_path only for relative paths and neither rejects absolute paths nor canonicalizes joined paths before enforcing workspace containment. As a result, tool arguments or model-generated function calls to grep_search, glob_search, read_file, or list_directory can supply absolute paths or '../' traversal sequences to read, search, and enumerate files outside the intended workspace directory, with file contents returned to the caller or injected into the model's tool-result context.

Phalcon is a high-performance, full-stack PHP framework. Prior to 5.14.1, Phalcon\Encryption\Crypt::decrypt compares the attacker-supplied HMAC tag against the freshly computed HMAC using PHP/Zephir identity comparison, which lowers to a byte-wise comparison that returns early on the first differing byte. This observable timing discrepancy can allow an attacker to recover a valid tag byte-by-byte and attach it to a chosen IV and ciphertext so that decrypt() accepts tampered encrypted content as authentic. This issue is fixed in version 5.14.1.

Deloitte AI Assist for Customer disclosed some configuration information through public-facing API endpoints that accepted unauthenticated requests. This information could reduce an attacker’s reconnaissance effort. On 2026-03-25, AI Assist for Customer restricted network access and enforced authentication for the previously exposed endpoints.

Improper input validation in Samsung Email prior to version 6.2.13.1 allows local attackers to create arbitrary files within the application sandbox.

Software installed and run as a non-privileged user may conduct a sequence of improper GPU system calls causing use after free, which helps in facilitating unprivileged memory access from a shader code. Triggering failure path in the MMU mapping logic by a malicious code could lead to incomplete cleanup of an internal driver state, allowing for future unauthorized access to the contents of the physical memory.

LangChain4j is a Java library for building LLM-powered applications on the JVM. Prior to 1.2.1-beta8, 1.5.1-beta11, 1.11.8-beta19, and 1.16.3-beta26, the MariaDB and pgvector embedding stores build metadata-filter SQL by string-concatenating filter keys, and in MariaDB string values, directly into the query without adequate escaping. A crafted metadata key in EmbeddingSearchRequest.filter() can break out of its SQL context and inject arbitrary SQL into the statements executed by the stores' search and removeAll(Filter) operations, enabling blind data exfiltration, denial of service via sleep functions, and deletion of arbitrary rows through removeAll(Filter). This issue is fixed in langchain4j-mariadb and langchain4j-pgvector versions 1.2.1-beta8, 1.5.1-beta11, 1.11.8-beta19, and 1.16.3-beta26.