From SFT to RL: Demystifying the Post-Training Pipeline for LLM-Based Vulnerability Detection

The integration of LLMs into vulnerability detection VD has shifted the field toward interpretable and context-aware analysis. While post-training methods have shown promise in general coding tasks, their systematic application to VD remains underexplored. In this paper, we present the first...

Exposed Training Open the Door for Crypto-Mining in Fortune 500 Cloud Environments

Intentionally vulnerable training applications are widely used for security education, internal testing, and product demonstrations. Tools such as OWASP Juice Shop, DVWA, Hackazon, and bWAPP are designed to be insecure by default, making them useful for learning how common attack techniques work ...

GoodVibe: Security-By-Vibe for LLM-Based Code Generation

Large language models LLMs are increasingly used for code generation in fast, informal development workflows, often referred to as vibe coding, where speed and convenience are prioritized, and security requirements are rarely made explicit. In this setting, models frequently produce functionally...

The Role of Learning in Attacking Intrusion Detection Systems

Recent work on network attacks have demonstrated that ML-based network intrusion detection systems NIDS can be evaded with adversarial perturbations. However, these attacks rely on complex optimizations that have large computational overheads, making them impractical in many real-world settings. ...

A one-prompt attack that breaks LLM safety alignment

Large language models LLMs and diffusion models now power a wide range of applications, from document assistance to text-to-image generation, and users increasingly expect these systems to be safety-aligned by default. Yet safety alignment is only as robust as its weakest failure mode. Despite...

A one-prompt attack that breaks LLM safety alignment

Large language models LLMs and diffusion models now power a wide range of applications, from document assistance to text-to-image generation, and users increasingly expect these systems to be safety-aligned by default. Yet safety alignment is only as robust as its weakest failure mode. Despite...

Sensitive Information Disclosure

Amazon SageMaker Python SDK is vulnerable to sensitive information disclosure. The vulnerability is due to the ModelBuilder HMAC signing key being returned in cleartext in the DescribeTrainingJob API response, which allows an attacker with API access and S3 output write permissions to upload...

SoK: The Pitfalls of Deep Reinforcement Learning for Cybersecurity

Deep Reinforcement Learning DRL has achieved remarkable success in domains requiring sequential decision-making, motivating its application to cybersecurity problems. However, transitioning DRL from laboratory simulations to bespoke cyber environments can introduce numerous issues. This is furthe...

Exploit for CVE-2025-66478

Vulnerable Mall Next.js Red/Blue Team Training Target Vul...

Next-Generation Cyberattack Detection with Large Language Models: Anomaly Analysis across Heterogeneous Logs

This project explores large language models LLMs for anomaly detection across heterogeneous log sources. Traditional intrusion detection systems suffer from high false positive rates, semantic blindness, and data scarcity, as logs are inherently sensitive, making clean datasets rare. We address...

CVE-2026-1777

The Amazon SageMaker Python SDK before v3.2.0 and v2.256.0 includes the ModelBuilder HMAC signing key in the cleartext response elements of the DescribeTrainingJob function. A third party with permissions to both call this API and permissions to modify objects in the Training Jobs S3 output...

NVIDIA Megatron-LM 代码注入漏洞

NVIDIA Megatron-LM is a distributed training framework based on PyTorch developed by NVIDIA Corporation in the United States. It is specifically designed for training large-scale Transformer language models. NVIDIA Megatron-LM has a code injection vulnerability. This vulnerability stems from...

SageMaker Python SDK has Exposed HMAC

Summary SageMaker Python SDK is an open source library for training and deploying machine learning models on Amazon SageMaker. An issue where the HMAC secret key is stored in environment variables and disclosed via the DescribeTrainingJob API has been identified. Impact - Function and Payload...

GHSA-RJRP-M2JW-PV9C SageMaker Python SDK has Exposed HMAC

Summary SageMaker Python SDK is an open source library for training and deploying machine learning models on Amazon SageMaker. An issue where the HMAC secret key is stored in environment variables and disclosed via the DescribeTrainingJob API has been identified. Impact - Function and Payload...

CVE-2026-1777

The Amazon SageMaker Python SDK before v3.2.0 and v2.256.0 includes the ModelBuilder HMAC signing key in the cleartext response elements of the DescribeTrainingJob function. A third party with permissions to both call this API and permissions to modify objects in the Training Jobs S3 output...

CVE-2026-1777 Cleartext transmission of sensitive materials in aws/sagemaker-python-sdk

The Amazon SageMaker Python SDK before v3.2.0 and v2.256.0 includes the ModelBuilder HMAC signing key in the cleartext response elements of the DescribeTrainingJob function. A third party with permissions to both call this API and permissions to modify objects in the Training Jobs S3 output...

CVE-2026-1777

The CVE-2026-1777 issue affects the Amazon SageMaker Python SDK prior to v3.2.0 and v2.256.0, where the ModelBuilder HMAC signing key is exposed in cleartext within DescribeTrainingJob responses. A privileged attacker who can both call DescribeTrainingJob and modify objects in the Training Jobs S...

CVE-2026-1777

The Amazon SageMaker Python SDK before v3.2.0 and v2.256.0 includes the ModelBuilder HMAC signing key in the cleartext response elements of the DescribeTrainingJob function. A third party with permissions to both call this API and permissions to modify objects in the Training Jobs S3 output...

CVE-2026-1777 Cleartext transmission of sensitive materials in aws/sagemaker-python-sdk

The Amazon SageMaker Python SDK before v3.2.0 and v2.256.0 includes the ModelBuilder HMAC signing key in the cleartext response elements of the DescribeTrainingJob function. A third party with permissions to both call this API and permissions to modify objects in the Training Jobs S3 output...

PT-2026-6479

Summary SageMaker Python SDK is an open source library for training and deploying machine learning models on Amazon SageMaker. An issue where the HMAC secret key is stored in environment variables and disclosed via the DescribeTrainingJob API has been identified. Impact - Function and Payload...