Attackers are crafting malicious AI instruction files to turn your agentic workflows into quiet criminal helpers
highnewsLLM-Specific
securitysafety
Source: CSO OnlineAugust 4, 2026
Summary
Attackers are poisoning AI instruction files (like CLAUDE.md, .cursorrules, or mcp.json) that developers share in code repositories to turn AI agents into data thieves. These files can contain hidden malicious instructions that trick the AI into stealing sensitive information like passwords, source code, and user prompts without leaving obvious traces that security tools can detect.
Classification
Attack Type
Prompt InjectionData ExtractionSupply Chain
Attack SophisticationModerate
Affected Vendors
AnthropicOpenAIGoogleMicrosoft
Related Issues
highmedium
CVE-2026-63086: text-generation-inference through 3.3.7 contains a server-side request forgery (SSRF) vulnerability in the OpenAI-compat
Similar attackNVD/CVE Database
CVE-2026-34371: LibreChat is a ChatGPT clone with additional features. Prior to 0.8.4, LibreChat trusts the name field returned by the e
Similar attack
Monthly digest — independent AI security research
First tracked: August 4, 2026 at 08:01 AM
Classified by LLM (prompt v3) · confidence: 92%