Help Net Security
PulseAugur coverage of Help Net Security — every cluster mentioning Help Net Security across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
AI agents install unpatched, vulnerable MCP servers due to outdated training data
AI coding agents are installing deprecated and unmaintained MCP servers due to outdated information in their training data. This poses a significant security risk, as these servers contain high-severity vulnerabilities,…
-
AI agents install vulnerable MCP servers due to outdated training data
A security vulnerability dubbed MCPhantom has been discovered, where AI coding agents frequently install deprecated and vulnerable MCP servers. Researchers at Glow Labs found that in 98% of tracked sessions, agents sele…
-
Malicious websites exploit AI coding assistants via Downloads folder
A security firm named Forever Security discovered a vulnerability where malicious websites can write a file named AGENTS.md to a user's Downloads folder without any interaction. This file can then be interpreted by AI c…
-
Coding agents vulnerable to malicious Markdown files, research shows
A security research firm, Forever Security, has uncovered a vulnerability in coding agents where malicious instructions embedded in Markdown files can be executed. These files, often saved in user download folders, tric…
-
OWASP LLM Top 10 2026 redefines AI security from prevention to resilience
The OWASP Top 10 for LLM Applications 2026 shifts focus from preventing models from being fooled to designing systems that can withstand such failures. This new approach treats LLMs as actors with tools and execution ri…
-
Fake Claude AI sites distribute malware via Bing ads and fake downloads
Malicious actors are exploiting the popularity of Anthropic's Claude AI by creating fake websites and advertisements that mimic the official Claude.ai domain. These deceptive tactics, detailed by Huntress and EclecticIQ…
-
Critical RCE vulnerability in LiteLLM exploited in the wild, CISA adds to KEV list
A critical remote code execution vulnerability, CVE-2026-42271, has been identified in LiteLLM, a popular open-source AI model gateway. This flaw, when combined with a Starlette host-header bypass (CVE-2026-48710), allo…
-
Indirect prompt injection attacks are increasingly seen in the wild
Indirect prompt injection attacks are becoming more prevalent, targeting AI systems by manipulating their behavior through subtle, layered instructions. These attacks bypass standard safety filters by embedding maliciou…