OWASP
PulseAugur coverage of OWASP — every cluster mentioning OWASP across labs, papers, and developer communities, ranked by signal.
- 2026-05-13 research_milestone OWASP added memory poisoning as a critical vulnerability (ASI06) to its Top 10 for Agentic Applications. source
2 day(s) with sentiment data
-
OWASP releases Cornucopia game for security team training
OWASP has released Cornucopia, a game designed to help teams improve their application security processes. The game focuses on threat modeling and requirement analysis, encouraging collaborative learning and growth with…
-
AI agents vulnerable to memory poisoning attacks, OWASP warns
A new security vulnerability, termed memory poisoning, has been identified in AI agents that utilize persistent memory stores. This attack allows malicious actors to inject false information into an agent's memory, caus…
-
Anthropic's Claude 4.7, Qwen Image 2.0, and Serverless GPUs highlighted
This TLDR AI newsletter covers several AI developments, including Anthropic's Claude 4.7 model, Alibaba's Qwen Image 2.0, and advancements in serverless GPUs. It also promotes a SANS eBook on an AI Security Maturity Model.
-
OWASP releases Copi game engine for AI threat modeling
OWASP has released Copi, a free game engine designed to help teams conduct threat modeling. The new Cornucopia Companion Edition v1.0 includes six suits covering Agentic AI, Automated Threats, Cloud, Frontend, Large Lan…
-
OWASP releases anniversary edition of threat modeling card game
OWASP has released the 25th anniversary edition of its threat modeling card game, Cornucopia. This game is designed to help software development teams identify security requirements within agile development processes. B…
-
OWASP Top 10 list details LLM security risks
The OWASP Top 10 for LLM Applications (2025) identifies critical security risks for AI-powered systems, extending beyond traditional vulnerabilities due to LLMs' interaction with prompts, data, and tools. Key risks incl…
-
Prompt injection is the top LLM security risk in 2026
Prompt injection remains the primary security threat for LLM applications in 2026, as identified by OWASP LLM01. Attackers can exploit this vulnerability to steal data, bypass safety measures, or perform unauthorized ac…
-
AI agents have two souls: controllable code and unpredictable LLMs
AI agents are defined by their ability to interpret inputs, reason, and take actions using generative AI models. A preferred technical definition, consistent with OWASP, outlines five core components: the generative AI …
-
MLOps security benchmark maps OWASP and MITRE ATLAS to ML pipelines
This article details the creation of an MLOps security benchmark by mapping the OWASP Top 10 for LLMs and the MITRE ATLAS framework onto a practical machine learning pipeline. The author outlines the process of integrat…
-
Prompt injection defenses focus on structural safeguards, not model intelligence
This article outlines six patterns for defending against prompt injection attacks in large language models, emphasizing that defenses should not rely on the model's inherent intelligence. The author proposes implementin…
-
Prompt injection defenses focus on structural safeguards, not model intelligence
This article outlines six patterns for defending against prompt injection attacks in large language models, emphasizing that defenses should not rely on the model's inherent intelligence. The author proposes implementin…
-
Bawbel Scanner v1.1.0 adds attack chain, server-card, and rug pull detection for AI agents
The Bawbel Scanner has been updated to version 1.1.0, introducing enhanced security features for agentic AI components. This release includes toxic flow detection to identify combined attack chains, server-card scanning…
-
OWASP Standard Guides AI Use in Penetration Testing
The OWASP Penetration Testing Standard is being updated to include guidelines for the ethical and effective use of artificial intelligence in penetration testing. This initiative aims to provide a governance framework f…
-
Vibe-coded SaaS applications flagged for security vulnerabilities by OWASP ZAP
A user on Mastodon highlighted a potential security vulnerability in "vibe-coded" SaaS applications, suggesting they are prone to issues identified by OWASP ZAP reports. The post implies that applications developed with…
-
Sigmoid.social's measurement work aligns with Fortune's AI cybersecurity standards piece
A quote from Mastodon user sigmoid.social, attributed to BIML, appeared in a Fortune article discussing AI cybersecurity standards. The user noted that their upcoming work on measurement aligns well with the article's timing.
-
38% of MCP servers have no auth -- inside the OWASP MCP Top 10
A new open-source project, Claw Code, has been released, offering a Rust implementation for an agent CLI harness that can interact with models like Anthropic's Claude. The project emphasizes building from source and pro…