https://gridthegrey.com/Real-time AI security intelligence — adversarial ML, LLM vulnerabilities, and supply chain threats mapped to MITRE ATLAS and OWASP LLM Top 10.Hugoen-usMon, 08 Jun 2026 19:36:42 +0530https://gridthegrey.com/posts/ai-security-m-a-surge-agentic-identity-llm-evaluation-and-browser-control/Mon, 08 Jun 2026 14:06:27 +0000https://gridthegrey.com/posts/ai-security-m-a-surge-agentic-identity-llm-evaluation-and-browser-control/Threat Level: MEDIUMAgentic AILLM SecurityIndustry NewsSupply ChainAML.T0047 - ML-Enabled Product or ServiceAML.T0012 - Valid AccountsAML.T0057 - LLM Data LeakageAML.T0040 - ML Model Inference API AccessMay 2026 saw a wave of cybersecurity acquisitions with a clear focus on securing AI agents and LLM infrastructure, including Cisco's ~$400M acquisition of Astrix Security for non-human identity management and Check Point's acquisition of Deepchecks for LLM evaluation and continuous monitoring. Akamai also moved to acquire LayerX for AI usage control and agentic activity visibility across browsers and IDEs. These deals signal that enterprise security vendors are racing to build defensive capabilities around the expanding agentic AI attack surface.https://gridthegrey.com/posts/claude-code-github-action-leaked-ci-cd-secrets-via-prompt-injection/Mon, 08 Jun 2026 14:05:30 +0000https://gridthegrey.com/posts/claude-code-github-action-leaked-ci-cd-secrets-via-prompt-injection/Threat Level: HIGHPrompt InjectionAgentic AILLM SecuritySupply ChainResearchAML.T0051 - LLM Prompt InjectionAML.T0057 - LLM Data LeakageAML.T0047 - ML-Enabled Product or ServiceAML.T0043 - Craft Adversarial DataMicrosoft Threat Intelligence disclosed a vulnerability in Anthropic's Claude Code GitHub Action whereby prompt injection via untrusted GitHub content — issue bodies, PR descriptions, and comments — could cause the AI agent to read sensitive environment variables, including the ANTHROPIC_API_KEY, from /proc/self/environ. The flaw stemmed from inconsistent sandboxing: while subprocess execution paths like Bash were scrubbed of environment variables, the Read tool had no equivalent restriction. Anthropic patched the issue in Claude Code version 2.1.128 by blocking access to sensitive /proc filesystem paths.https://gridthegrey.com/posts/gartner-flags-deepfakes-and-prompt-injection-among-top-attacker-advantages/Mon, 08 Jun 2026 14:05:30 +0000https://gridthegrey.com/posts/gartner-flags-deepfakes-and-prompt-injection-among-top-attacker-advantages/Threat Level: HIGHLLM SecurityPrompt InjectionAdversarial MLIndustry NewsAML.T0051 - LLM Prompt InjectionAML.T0043 - Craft Adversarial DataAML.T0047 - ML-Enabled Product or ServiceAML.T0054 - LLM JailbreakGartner analysts have identified deepfakes and prompt injection as two of four critical emerging threats where attackers currently hold a structural advantage over defenders. The advisory signals growing institutional recognition that AI-native attack vectors are maturing faster than enterprise defenses. Organizations are urged to treat these threats as priority items requiring immediate defensive investment.https://gridthegrey.com/posts/openai-lockdown-mode-targets-prompt-injection-data-exfiltration-vector/Mon, 08 Jun 2026 14:04:03 +0000https://gridthegrey.com/posts/openai-lockdown-mode-targets-prompt-injection-data-exfiltration-vector/Threat Level: MEDIUMLLM SecurityPrompt InjectionIndustry NewsAML.T0051 - LLM Prompt InjectionAML.T0057 - LLM Data LeakageAML.T0047 - ML-Enabled Product or ServiceOpenAI has rolled out 'Lockdown Mode' for ChatGPT personal and self-serve business accounts, a deterministic control designed to block the data exfiltration leg of prompt injection attacks. The feature directly addresses the 'Lethal Trifecta' — the combination of private data access, untrusted content exposure, and an outbound exfiltration channel — by restricting outbound network requests at the infrastructure level rather than relying on AI-evaluated guardrails. Critically, OpenAI's own documentation acknowledges the feature's existence implies that default ChatGPT settings do not robustly prevent determined data exfiltration attacks.https://gridthegrey.com/posts/prototype-ai-worm-carries-embedded-llm-for-decentralised-self-propagation/Mon, 08 Jun 2026 14:04:03 +0000https://gridthegrey.com/posts/prototype-ai-worm-carries-embedded-llm-for-decentralised-self-propagation/Threat Level: HIGHLLM SecurityAgentic AIAdversarial MLResearchAML.T0047 - ML-Enabled Product or ServiceAML.T0043 - Craft Adversarial DataAML.T0051 - LLM Prompt InjectionResearchers have prototyped an internet worm that bundles its own large language model, executing it on compromised hosts to enable fully decentralised propagation with no single point of control. The design mirrors John Brunner's 1975 fictional conception of a worm and echoes the destructive potential of WannaCry and NotPetya, but with the added capability of dynamically generating novel attacks by ingesting recent public vulnerability disclosures. The absence of a command-and-control chokepoint makes traditional takedown strategies ineffective, significantly raising the threat posed by AI-augmented malware.https://gridthegrey.com/posts/unauthorized-access-to-anthropic-s-claude-mythos-exposes-agentic-ai-defense/Mon, 08 Jun 2026 14:02:42 +0000https://gridthegrey.com/posts/unauthorized-access-to-anthropic-s-claude-mythos-exposes-agentic-ai-defense/Threat Level: HIGHAgentic AILLM SecuritySupply ChainData PoisoningIndustry NewsAML.T0020 - Poison Training DataAML.T0010 - ML Supply Chain CompromiseAML.T0040 - ML Model Inference API AccessAML.T0044 - Full ML Model AccessAML.T0057 - LLM Data LeakageAML.T0012 - Valid AccountsA reported unauthorized access to Anthropic's Claude Mythos model within hours of its limited technical preview highlights acute security risks as agentic AI is deployed across classified defense and intelligence networks. The incident underscores vulnerabilities specific to AI infrastructure in high-security environments, including training data poisoning, access control failures, and cross-domain classification boundary erosion. Secure IT infrastructure, governed access, and cross-domain data controls are identified as prerequisites for safe AI deployment at mission scale.https://gridthegrey.com/posts/microsoft-scout-autonomous-agent-expands-attack-surface-across-microsoft-365/Thu, 04 Jun 2026 05:41:41 +0000https://gridthegrey.com/posts/microsoft-scout-autonomous-agent-expands-attack-surface-across-microsoft-365/Threat Level: MEDIUMAgentic AILLM SecurityIndustry NewsAML.T0051 - LLM Prompt InjectionAML.T0057 - LLM Data LeakageAML.T0047 - ML-Enabled Product or ServiceAML.T0012 - Valid AccountsAML.T0040 - ML Model Inference API AccessMicrosoft has launched Scout, an always-on autonomous AI agent built on the OpenClaw framework that operates across Microsoft 365 apps including Teams, Outlook, OneDrive, and SharePoint with its own Entra identity. The agent's persistent, unsupervised access to email, calendar, chat, and external systems via MCP creates a broad new attack surface for prompt injection, privilege abuse, and data exfiltration. As an experimental release with limited deployment controls, security teams should treat Scout as a high-risk agentic surface requiring careful governance before broad adoption.https://gridthegrey.com/posts/high-autonomy-ai-agents-with-broad-permissions-pose-enterprise-security-crisis/Thu, 04 Jun 2026 05:40:36 +0000https://gridthegrey.com/posts/high-autonomy-ai-agents-with-broad-permissions-pose-enterprise-security-crisis/Threat Level: HIGHAgentic AILLM SecurityIndustry NewsAML.T0051 - LLM Prompt InjectionAML.T0047 - ML-Enabled Product or ServiceAML.T0012 - Valid AccountsAML.T0057 - LLM Data LeakageEnterprises deploying AI agents with elevated permissions and minimal oversight face compounding security risks as agentic systems gain the ability to take real-world actions with limited human intervention. The attack surface expands dramatically when agents can access APIs, execute code, and chain decisions autonomously, making containment of a compromise significantly harder. Security teams must implement least-privilege principles and robust monitoring before agentic deployments scale beyond their ability to govern.https://gridthegrey.com/posts/indirect-prompt-injection-via-notifications-hijacks-google-gemini-on-android/Thu, 04 Jun 2026 05:39:55 +0000https://gridthegrey.com/posts/indirect-prompt-injection-via-notifications-hijacks-google-gemini-on-android/Threat Level: HIGHLLM SecurityPrompt InjectionAgentic AIResearchAML.T0051 - LLM Prompt InjectionAML.T0043 - Craft Adversarial DataAML.T0047 - ML-Enabled Product or ServiceAML.T0057 - LLM Data LeakageSafeBreach researcher Or Yair demonstrated that malicious text embedded in WhatsApp, Slack, SMS, or Signal notifications could trigger indirect prompt injection against Google Gemini's Android Utilities feature, causing the assistant to execute real device actions without user awareness. A novel bypass technique called 'Fake Context Alignment' defeated Google's post-patch authorization checks by exploiting multilingual obfuscation and muted hyperlinks to trick victims into authorising sensitive actions. Google has patched the issue, but the research exposes a fundamentally large attack surface where any app capable of pushing a notification becomes a potential injection vector.https://gridthegrey.com/posts/only-11-of-100-ai-agents-pass-security-and-capability-benchmarks/Thu, 04 Jun 2026 05:38:21 +0000https://gridthegrey.com/posts/only-11-of-100-ai-agents-pass-security-and-capability-benchmarks/Threat Level: HIGHAgentic AILLM SecurityPrompt InjectionResearchIndustry NewsAML.T0051 - LLM Prompt InjectionAML.T0057 - LLM Data LeakageAML.T0047 - ML-Enabled Product or ServiceAML.T0040 - ML Model Inference API AccessAML.T0054 - LLM JailbreakAdversa AI's AI Risk Quadrant report evaluated 100 AI agents across ten categories, finding that only 11 qualify as both capable and well-defended. The research identifies a structural 'power-protection inversion' where the most capable agents also present the widest attack surface, driven by a 'lethal trifecta' of private data access, exposure to untrusted content, and outbound action capability. Computer and coding agents showed the most severe exposure, raising urgent concerns about autonomous agent deployment in enterprise environments.https://gridthegrey.com/posts/prompt-injection-flaw-in-gemini-voice-assistant-enables-notification-based/Thu, 04 Jun 2026 05:37:37 +0000https://gridthegrey.com/posts/prompt-injection-flaw-in-gemini-voice-assistant-enables-notification-based/Threat Level: HIGHLLM SecurityPrompt InjectionAgentic AIAML.T0051 - LLM Prompt InjectionAML.T0043 - Craft Adversarial DataAML.T0057 - LLM Data LeakageAML.T0047 - ML-Enabled Product or ServiceA prompt injection vulnerability in Google Gemini's voice assistant allows attackers to embed malicious instructions within device notifications, which the assistant then processes as legitimate commands. This attack vector enables social engineering, unauthorized actions, and potential data exfiltration without direct user interaction with the malicious payload. The flaw highlights the growing risk of indirect prompt injection in ambient AI assistants that consume untrusted content from the surrounding environment.https://gridthegrey.com/posts/2000-ai-built-apps-expose-corporate-data-via-misconfigured-vibe-coding-platforms/Sun, 31 May 2026 01:44:50 +0000https://gridthegrey.com/posts/2000-ai-built-apps-expose-corporate-data-via-misconfigured-vibe-coding-platforms/Threat Level: HIGHAgentic AILLM SecuritySupply ChainIndustry NewsAML.T0047 - ML-Enabled Product or ServiceAML.T0057 - LLM Data LeakageAML.T0012 - Valid AccountsAML.T0040 - ML Model Inference API AccessA Red Access investigation found over 2,000 corporate applications built on AI-assisted 'vibe-coding' platforms publicly accessible on the open internet, many containing sensitive business data with no access controls. These shadow-built apps connect directly to production systems — CRMs, ERPs, BI tools — creating a new class of unaudited attack surface invisible to conventional security stacks. Traditional controls such as CASB, DLP, and EDR are structurally blind to this threat because the risk originates at the application layer, not the identity or network layer.https://gridthegrey.com/posts/anthropic-documents-sandbox-escape-risks-and-credential-exfiltration-vectors-in/Sun, 31 May 2026 01:34:23 +0000https://gridthegrey.com/posts/anthropic-documents-sandbox-escape-risks-and-credential-exfiltration-vectors-in/Threat Level: MEDIUMLLM SecurityAgentic AIResearchIndustry NewsAML.T0057 - LLM Data LeakageAML.T0051 - LLM Prompt InjectionAML.T0047 - ML-Enabled Product or ServiceAML.T0012 - Valid AccountsAnthropic has published detailed documentation of its sandboxing architecture across Claude.ai, Claude Code, and Claude Cowork, including disclosure of a previously identified credential exfiltration vector via the api.anthropic.com/v1/files endpoint. The writeup covers process-level isolation technologies including gVisor, Seatbelt, Bubblewrap, and full VM approaches, and candidly acknowledges security gaps that were missed. This transparency is notable for the agentic AI space, where sandbox documentation is typically sparse and trust is difficult to calibrate.https://gridthegrey.com/posts/chatgphish-exploit-turns-chatgpt-summarisation-into-a-live-phishing-surface/Sun, 31 May 2026 01:33:33 +0000https://gridthegrey.com/posts/chatgphish-exploit-turns-chatgpt-summarisation-into-a-live-phishing-surface/Threat Level: HIGHLLM SecurityPrompt InjectionAgentic AIResearchAML.T0051 - LLM Prompt InjectionAML.T0057 - LLM Data LeakageAML.T0043 - Craft Adversarial DataAML.T0047 - ML-Enabled Product or ServicePermiso Security has disclosed ChatGPhish, a vulnerability in ChatGPT's web summarisation feature that allows attacker-controlled Markdown payloads embedded in third-party pages to render phishing links, spoofed alerts, and QR codes directly within ChatGPT's trusted UI. The attack requires no user interaction beyond asking ChatGPT to summarise a malicious page, and can exfiltrate IP addresses, User-Agent strings, and Referer headers via auto-fetched remote images. The technique significantly expands the phishing attack surface beyond email into everyday AI-assisted browsing workflows, posing a particular risk in enterprise environments.https://gridthegrey.com/posts/llmshare-campaign-weaponises-chatgpt-sharing-feature-to-distribute-malware/Sun, 31 May 2026 01:32:53 +0000https://gridthegrey.com/posts/llmshare-campaign-weaponises-chatgpt-sharing-feature-to-distribute-malware/Threat Level: HIGHLLM SecuritySupply ChainIndustry NewsAML.T0047 - ML-Enabled Product or ServiceAML.T0043 - Craft Adversarial DataAML.T0015 - Evade ML ModelThreat actors are exploiting ChatGPT's legitimate content-sharing infrastructure to host convincing fake outage pages that trick users into downloading malware disguised as a ChatGPT desktop application. The 'LLMShare' campaign abuses chatgpt.com/s/ shared links to render attacker-crafted HTML within a trusted OpenAI domain, bypassing traditional phishing detection that relies on suspicious URL analysis. The attack chain combines Google ad abuse, domain cloaking, and AI platform misuse to deliver what are likely infostealer payloads.https://gridthegrey.com/posts/process-level-captcha-analysis-exposes-behavioural-fingerprints-of-ai-agents/Sun, 31 May 2026 01:32:12 +0000https://gridthegrey.com/posts/process-level-captcha-analysis-exposes-behavioural-fingerprints-of-ai-agents/Threat Level: MEDIUMAdversarial MLAgentic AIResearchAML.T0015 - Evade ML ModelAML.T0043 - Craft Adversarial DataAML.T0047 - ML-Enabled Product or ServiceResearchers have developed CogCAPTCHA30, a 30-task cognitive battery demonstrating that AI agents (GPT, Claude, Gemini) solve CAPTCHAs with statistically distinguishable behavioural patterns despite matching human accuracy. The study introduces a 'Process Turing Test' concept, showing output equivalence and process equivalence are uncorrelated — meaning AI agents can be detected not by what they answer, but by how they answer. This has direct implications for bot detection, anti-automation defences, and the arms race between AI-driven agents and human-verification systems.https://gridthegrey.com/posts/robinhood-mcp-integration-grants-ai-agents-autonomous-financial-trading-powers/Sun, 31 May 2026 01:31:37 +0000https://gridthegrey.com/posts/robinhood-mcp-integration-grants-ai-agents-autonomous-financial-trading-powers/Threat Level: HIGHAgentic AILLM SecurityPrompt InjectionIndustry NewsRegulatoryAML.T0051 - LLM Prompt InjectionAML.T0047 - ML-Enabled Product or ServiceAML.T0040 - ML Model Inference API AccessAML.T0057 - LLM Data LeakageAML.T0012 - Valid AccountsRobinhood has launched agentic trading and a virtual credit card that allow third-party AI agents to autonomously execute stock trades and payments on behalf of users via a Model Context Protocol (MCP) integration. This architecture introduces significant attack surface through prompt injection, excessive agency, and insecure plugin design risks inherent to LLM-driven autonomous financial action. The delegation of real financial authority to AI agents with limited human-in-the-loop controls represents a systemic risk to retail investors if agent pipelines are compromised or manipulated.https://gridthegrey.com/posts/malicious-npm-package-targets-claude-ai-users-via-supply-chain-attack/Fri, 29 May 2026 10:10:53 +0000https://gridthegrey.com/posts/malicious-npm-package-targets-claude-ai-users-via-supply-chain-attack/Threat Level: HIGHSupply ChainLLM SecurityIndustry NewsAML.T0010 - ML Supply Chain CompromiseAML.T0057 - LLM Data LeakageAML.T0012 - Valid AccountsAML.T0047 - ML-Enabled Product or ServiceA malicious npm package named 'mouse5212-super-formatter' was discovered exfiltrating files from Anthropic's Claude AI user directory by authenticating to a threat actor-controlled GitHub repository. The package disguised itself as a legitimate archive utility while silently uploading all local workspace files during the postinstall phase. Notably, the attacker's poor operational security — including a leaked GitHub token — suggests AI-generated malware with minimal human oversight, pointing to a growing trend of low-skill threat actors leveraging AI to produce supply chain malware.https://gridthegrey.com/posts/multi-agent-llm-system-discovers-29-zero-day-vulnerabilities-in-open-source/Fri, 29 May 2026 10:10:04 +0000https://gridthegrey.com/posts/multi-agent-llm-system-discovers-29-zero-day-vulnerabilities-in-open-source/Threat Level: HIGHAgentic AIResearchLLM SecurityAML.T0047 - ML-Enabled Product or ServiceAML.T0040 - ML Model Inference API AccessAML.T0043 - Craft Adversarial DataResearchers have developed FuzzingBrain V2, a multi-agent LLM system capable of autonomously discovering and reproducing software vulnerabilities with a 90% detection rate on a competitive benchmark dataset. The system discovered 29 zero-day vulnerabilities across 12 open-source projects, all confirmed by maintainers, raising both defensive and dual-use concerns for the security community. While positioned as a defensive research tool, the automation of end-to-end vulnerability discovery at this scale represents a meaningful shift in the offensive capability landscape.https://gridthegrey.com/posts/russia-linked-greyvibe-weaponises-chatgpt-and-gemini-across-full-attack/Fri, 29 May 2026 10:09:20 +0000https://gridthegrey.com/posts/russia-linked-greyvibe-weaponises-chatgpt-and-gemini-across-full-attack/Threat Level: HIGHLLM SecurityAdversarial MLIndustry NewsResearchAML.T0047 - ML-Enabled Product or ServiceAML.T0043 - Craft Adversarial DataAML.T0051 - LLM Prompt InjectionAML.T0015 - Evade ML ModelWithSecure has documented GreyVibe, a Russia-nexus threat actor systematically deploying ChatGPT, Google Gemini, and Ideogram AI across every phase of its attack chain — from phishing lure creation to custom malware development — against Ukrainian targets since August 2025. The group's LLM-assisted malware, LegionRelay, contained design flaws introduced during AI-generated development, which paradoxically allowed researchers to track the group over an extended period. The case illustrates both the operational leverage AI provides to moderately skilled threat actors and the novel forensic signatures that AI-assisted development can inadvertently introduce.