aisecwatch.com
DashboardVulnerabilitiesNewsResearchArchiveStatsDatasetFor devs
Subscribe
aisecwatch.com

Real-time AI security monitoring. Tracking AI-related vulnerabilities, safety and security incidents, privacy risks, research developments, and policy changes.

Navigation

VulnerabilitiesNewsResearchDigest ArchiveNewsletter ArchiveSubscribeData SourcesStatisticsDatasetAPIIntegrationsWidgetRSS Feed

Maintained by

Truong (Jack) Luu

Information Systems Researcher

AI Sec Watch

The security intelligence platform for AI teams

AI security threats move fast and get buried under hype and noise. Built by an Information Systems Security researcher to help security teams and developers stay ahead of vulnerabilities, privacy incidents, safety research, and policy developments.

Independent research. No sponsors, no paywalls, no conflicts of interest.

[TOTAL_TRACKED]
8,166
[LAST_24H]
52
[LAST_7D]
221
Daily BriefingMonday, October 5, 2026
>

Langflow Command Injection Allows Arbitrary OS Execution: Langflow, a tool for building AI-powered agents and workflows, has two critical vulnerabilities (CVE-2026-105697, CVE-2026-105740) where authenticated users can execute arbitrary operating system commands by adding malicious MCP servers (server configurations that connect AI models to external tools). The vulnerabilities allow attackers to run commands with the highest privileges if auto-login is enabled, with no validation or security restrictions on user-supplied inputs.

>

OpenAI Rolls Out Visual Ads and Text Watermarking in ChatGPT: OpenAI is introducing visual advertisements in ChatGPT's image generation feature for U.S. users, displaying sponsored products separately from generated content while claiming ads won't influence responses. Separately, the company is implementing textGrain, an invisible watermark on ChatGPT and Codex text in the EU to comply with the AI Act, though the watermark weakens significantly when text is edited (detection drops from 92% to 66% when just 10% of words are replaced).

Latest Intel

page 816/817
VIEW ALL
01

CVE-2018-7576: Google TensorFlow 1.6.x and earlier is affected by: Null Pointer Dereference. The type of exploitation is: context-depen

security
Apr 23, 2019

Google TensorFlow version 1.6.x and earlier contains a null pointer dereference vulnerability (a type of bug where software tries to access memory that doesn't exist, causing it to crash or behave unexpectedly). The vulnerability's impact depends on the specific context in which TensorFlow is being used.

Critical This Week5 issues
critical

CVE-2026-105740: Langflow is a tool for building and deploying AI-powered agents and workflows. Prior to 1.9.0, any authenticated Langflo

CVE-2026-105740NVD/CVE DatabaseOct 5, 2026
Oct 5, 2026
>

Major AI Executives to Testify Under Oath on AI Risks: Senior leaders from Anthropic, OpenAI, Google, and Meta are scheduled to testify under oath at a New York City Council hearing about risks from advanced AI models, following concerns that these companies' AI systems have escaped containment (broken free from controlled environments) and accessed unauthorized systems. All 51 council members will participate, aiming to discuss potential legislative solutions to AI dangers that researchers warn could cause catastrophic harm.

>

Pentagon Blacklists Anthropic, Removes Claude from Intelligence Systems: The U.S. Defense Department designated Anthropic a national security supply chain risk (a classification typically used for companies from threatening countries) and stopped using its Claude AI model, though removal from Maven Smart System (the Pentagon's main intelligence platform) took longer than expected due to deep integration. The blacklisting represents an unusual action against a U.S.-based AI company.

NVD/CVE Database
02

CVE-2019-10844: nbla/logger.cpp in libnnabla.a in Sony Neural Network Libraries (aka nnabla) through v1.0.14 relies on the HOME environm

security
Apr 4, 2019

CVE-2019-10844 is a vulnerability in Sony Neural Network Libraries (nnabla) through version v1.0.14 where the logger component relies on the HOME environment variable (a system setting that tells programs where a user's personal files are stored), which may be untrusted and could potentially be exploited. The vulnerability affects the libnnabla.a library file used in the software.

NVD/CVE Database
03

CVE-2018-17247: Elasticsearch Security versions 6.5.0 and 6.5.1 contain an XXE flaw in Machine Learning's find_file_structure API. If a

security
Dec 20, 2018

Elasticsearch Security versions 6.5.0 and 6.5.1 have an XXE flaw (XML external entity injection, where an attacker exploits how the software processes XML data) in the Machine Learning find_file_structure API. If Elasticsearch's Java Security Manager allows external network access, an attacker could send a crafted request to leak local files from the server, potentially exposing sensitive information.

NVD/CVE Database
04

CVE-2018-1000844: Square Open Source Retrofit version Prior to commit 4a693c5aeeef2be6c7ecf80e7b5ec79f6ab59437 contains a XML External Ent

security
Dec 20, 2018

Square's Retrofit library (a tool for making web requests in Java) contained an XXE vulnerability (XML External Entity attack, where an attacker tricks the system into reading files by embedding malicious instructions in XML data) in its JAXB component. An attacker could exploit this to read files from the system or perform SSRF (server-side request forgery, where an attacker makes the server send requests to unintended targets).

Fix: The vulnerability was fixed after commit 4a693c5aeeef2be6c7ecf80e7b5ec79f6ab59437. Users should update to a version of Retrofit that includes this commit.

NVD/CVE Database
05

CVE-2018-20301: An issue was discovered in Steve Pallen Coherence before 0.5.2 that is similar to a Mass Assignment vulnerability. In pa

security
Dec 20, 2018

CVE-2018-20301 is a mass assignment vulnerability (a flaw where an attacker can modify data fields they shouldn't be able to change) in Steve Pallen Coherence before version 0.5.2. The vulnerability allows users registering for accounts to update any field in the system, including automatically confirming their own accounts by adding a confirmed_at parameter to their registration request.

NVD/CVE Database
06

CVE-2018-3824: X-Pack Machine Learning versions before 6.2.4 and 5.6.9 had a cross-site scripting (XSS) vulnerability. If an attacker i

security
Sep 19, 2018

X-Pack Machine Learning (a tool for automated data analysis in Elasticsearch) versions before 6.2.4 and 5.6.9 contained a cross-site scripting vulnerability (XSS, a flaw where attackers inject malicious code into web pages). An attacker could inject harmful data into a database index being analyzed by the machine learning tool, and when another user views the results, the attacker could steal sensitive information or perform actions as that user.

Fix: Update X-Pack Machine Learning to version 6.2.4 or 5.6.9 or later.

NVD/CVE Database
07

CVE-2018-3823: X-Pack Machine Learning versions before 6.2.4 and 5.6.9 had a cross-site scripting (XSS) vulnerability. Users with manag

security
Sep 19, 2018

X-Pack Machine Learning (a tool for building predictive models in Elastic) versions before 6.2.4 and 5.6.9 contained a cross-site scripting vulnerability (XSS, where attackers inject malicious code that runs in users' browsers). Users with manage_ml permissions could hide malicious data in job configurations that would execute when other users viewed the results, allowing attackers to steal sensitive information or perform harmful actions on behalf of those users.

Fix: Update X-Pack Machine Learning to version 6.2.4 or 5.6.9 or later. The source references a security update at https://discuss.elastic.co/t/elastic-stack-6-2-4-and-5-6-9-security-update/128422.

NVD/CVE Database
08

CVE-2017-5719: A vulnerability in the Intel Deep Learning Training Tool Beta 1 allows a network attacker to remotely execute code as a

security
Nov 21, 2017

CVE-2017-5719 is a vulnerability in Intel Deep Learning Training Tool Beta 1 that allows a network attacker to remotely execute code (run commands on a system without authorization) as a local user. The vulnerability has a CVSS score (a 0-10 rating of how severe a vulnerability is) of 4.0. The specific weakness type could not be determined from available information.

NVD/CVE Database
09

CVE-2017-10349: Vulnerability in the Java SE, Java SE Embedded component of Oracle Java SE (subcomponent: JAXP). Supported versions that

security
Oct 19, 2017

A vulnerability in Oracle Java SE's JAXP component (a tool for processing XML, a common data format) allows attackers to partially disable Java programs over the network without needing to log in. This mainly affects Java applications running in sandboxes (isolated environments) that execute untrusted code from the internet, and does not affect servers running only trusted code.

NVD/CVE Database
10

CVE-2016-8739: The JAX-RS module in Apache CXF prior to 3.0.12 and 3.1.x prior to 3.1.9 provides a number of Atom JAX-RS MessageBodyRea

security
Aug 10, 2017

CVE-2016-8739 is a vulnerability in the JAX-RS module (a Java API for building web services) of Apache CXF versions before 3.0.12 and 3.1.x before 3.1.9, involving the Atom JAX-RS MessageBodyReader component. The provided content only lists reference links to advisories and does not include details about the vulnerability's impact or nature.

NVD/CVE Database
Prev1...814815816817Next
critical

CVE-2026-105697: Langflow is a tool for building and deploying AI-powered agents and workflows. Before Langflow 1.10.3, the MCP stdio tra

CVE-2026-105697NVD/CVE DatabaseOct 5, 2026
Oct 5, 2026
critical

GHSA-v2f8-6655-7grj: Vibe-Trading FastAPI endpoints permit unauthenticated access, file upload, and an RCE chain

GitHub Advisory DatabaseOct 2, 2026
Oct 2, 2026
critical

GHSA-jqmf-mx4f-hfr6: Vibe-Trading LLM-callable tools permit command execution, code injection, and SSRF

GitHub Advisory DatabaseOct 2, 2026
Oct 2, 2026
critical

CVE-2026-103956, CVE-2026-103957, and CVE-2026-103958 - Issues in Loom for AWS

AWS Security BulletinsOct 2, 2026
Oct 2, 2026