PT-2025-27251

Name of the Vulnerable Software and Affected Versions: eosphoros-ai db-gpt versions up to 0.7.2 Description: A critical issue has been found, affecting the import flow function of the file /api/v2/serve/awel/flow/import. The manipulation of the File argument leads to path traversal, allowing for...

Security Assessment of DeepSeek and GPT Series Models against Jailbreak Attacks

The widespread deployment of large language models LLMs has raised critical concerns over their vulnerability to jailbreak attacks, i.e., adversarial prompts that bypass alignment mechanisms and elicit harmful or policy-violating outputs. While proprietary models like GPT-4 have undergone extensi...

Mechanistic Interpretability in the Presence of Architectural Obfuscation

Architectural obfuscation - e.g., permuting hidden-state tensors, linearly transforming embedding tables, or remapping tokens - has recently gained traction as a lightweight substitute for heavyweight cryptography in privacy-preserving large-language-model LLM inference. While recent work has sho...

LLM vs. SAST: a Technical Analysis on Detecting Coding Bugs of GPT4-Advanced Data Analysis

With the rapid advancements in Natural Language Processing NLP, large language models LLMs like GPT-4 have gained significant traction in diverse applications, including security vulnerability scanning. This paper investigates the efficacy of GPT-4 in identifying software vulnerabilities compared...

When GPT Spills the Tea: Comprehensive Assessment of Knowledge File Leakage in GPTs

Knowledge files have been widely used in large language model LLM agents, such as GPTs, to improve response quality. However, concerns about the potential leakage of knowledge files have grown significantly. Existing studies demonstrate that adversarial prompts can induce GPTs to leak knowledge...

airunner (>=3.0.0 <=3.1.7), athina (>=1.7.0 <=1.7.39) +26 more potentially affected by CVE-2025-1753 via llama-index-cli (>=0.1.13 <=0.4.0)

llama-index-cli PYPI version =0.1.13, =3.0.0, =1.7.0, =0.1.3, =0.1.169, =0.1.0, =0.3.0, =0.0.52, =1.0.9, =1.0.3.post1, =0.1.2, =0.1.7.dev20240924104148, =0.11.0, =0.11.23 - llama-index-callbacks-honeyhive =0.2.0 - llama-index-collection =0.2.0 and more Source cves: CVE-2025-1753 Source advisory:...

CVE-2024-10100

A path traversal vulnerability exists in binary-husky/gptacademic version 3.83. The vulnerability is due to improper handling of the file parameter, which is open to path traversal through URL encoding. This allows attackers to view any file on the host system, including sensitive files such as...

CVE-2024-54306

Cross-Site Request Forgery CSRF vulnerability in aitool AIKCT Engine Chatbot, ChatGPT, Gemini, GPT-4o Best AI Chatbot ai-seo-translator allows Cross Site Request Forgery.This issue affects AIKCT Engine Chatbot, ChatGPT, Gemini, GPT-4o Best AI Chatbot: from n/a through = 1.6.2...

CVE-2023-37273

Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. Running Auto-GPT version prior to 0.4.3 by cloning the git repo and executing docker compose run auto-gpt in the repo root uses a different docker-compose.yml file from the one suggested i...

CVE-2023-51528

Cross-Site Request Forgery CSRF vulnerability in Senol Sahin AI Power: Complete AI Pack – Powered by GPT-4.This issue affects AI Power: Complete AI Pack – Powered by GPT-4: from n/a through 1.8.12...

ACSE-Eval: Can LLMs Threat Model Real-World Cloud Infrastructure?

While Large Language Models have shown promise in cybersecurity applications, their effectiveness in identifying security threats within cloud deployments remains unexplored. This paper introduces AWS Cloud Security Engineering Eval, a novel dataset for evaluating LLMs cloud security threat...

Privacy-Preserving Transformers: SwiftKey'S Differential Privacy Implementation

In this paper we train a transformer using differential privacy DP for language modeling in SwiftKey. We run multiple experiments to balance the trade-off between the model size, run-time speed and accuracy. We show that we get small and consistent gains in the next-word-prediction and accuracy...

CVE-2025-4218

A vulnerability was found in handrew browserpilot up to 0.2.51. It has been declared as critical. Affected by this vulnerability is the function GPTSeleniumAgent of the file browserpilot/browserpilot/agents/gptseleniumagent.py. The manipulation of the argument instructions leads to code injection...

Enhancing Security and Strengthening Defenses in Automated Short-Answer Grading Systems

This study examines vulnerabilities in transformer-based automated short-answer grading systems used in medical education, with a focus on how these systems can be manipulated through adversarial gaming strategies. Our research identifies three main types of gaming strategies that exploit the...

PT-2025-18141

Name of the Vulnerable Software and Affected Versions huggingface/transformers library version v4.48.1 Description A Regular Expression Denial of Service ReDoS vulnerability was identified in the huggingface/transformers library, specifically in the file tokenization gpt neox japanese.py of the...

GPT Academic Open Redirect Vulnerability

GPT Academic is an interface that provides pragmatic interactions for LLM grand language models such as GPT/GLM. GPT Academic suffers from an open redirection vulnerability that originates from a user redirecting to a URL specified by the user-controlled file parameter without proper validation o...

GPT Academic Denial of Service Vulnerability (CNVD-2025-22736)

GPT Academic is an interface that provides pragmatic interactions for LLM grand language models such as GPT/GLM. GPT Academic suffers from a denial of service vulnerability that stems from the use of insecure regular expressions. An attacker could exploit this vulnerability to cause a regular...

GPT Academic path traversal vulnerability (CNVD-2025-22731)

GPT Academic is an interface that provides pragmatic interactions for LLM grand language models such as GPT/GLM. GPT Academic suffers from a path traversal vulnerability that can be exploited by an attacker to disclose sensitive information...

GPT Academic Path Traversal Vulnerability

GPT Academic is an interface that provides pragmatic interactions for LLM grand language models such as GPT/GLM. GPT Academic suffers from a path traversal vulnerability that can be exploited to cause an attacker to read a config.py file containing sensitive information...

GPT Academic Cross-Site Request Forgery Vulnerability (CNVD-2025-22738)

GPT Academic is an interface that provides pragmatic interactions for LLM grand language models such as GPT/GLM. GPT Academic suffers from a cross-site request forgery vulnerability that stems from a WEB application that does not adequately verify that a request is from a trusted user. An attacke...