LLM Security

258 reports

All LLM Security Agentic AI Industry News Research Supply Chain Prompt Injection First Look: Security Regulatory Jailbreaks Adversarial ML Data Poisoning Model Theft AI Security Tools Security Operations

Alibaba and Baidu Launch LLMs With US-Level Capabilities

FIRST LOOK ATLAS OWASP HIGH ▲ 6.2 Dark Reading Jul 04, 2026

Two newly released large language models from Chinese AI firms have reached capability parity with leading US frontier models, expanding the global pool of powerful AI available to both commercial and adversarial users. For defenders, this development broadens the asymmetry between attackers — who gain access to capable, potentially less-restricted models — and defenders, who must now account for threats generated by a wider set of model providers. Security teams should anticipate increased use of these models for offensive tasks such as phishing content generation, vulnerability research automation, and social engineering at scale.

Langflow LLM Agents Exploited for Ransomware Delivery

ATLAS OWASP CRITICAL ▲ 9.2 SecurityWeek Jul 04, 2026

A documented ransomware attack leveraged agentic AI infrastructure — specifically the Langflow LLM orchestration platform — to automate multi-stage intrusion chains combining known exploitation techniques with real-time LLM reasoning. This marks a significant escalation in threat actor capability, demonstrating that agentic AI can serve as an autonomous attack coordinator rather than merely an assistant. Security teams running self-hosted AI orchestration platforms now face an expanded attack surface where the AI layer itself can be both the entry point and the execution engine.

Microsoft Copilot MCP Tool Poisoning Enables Data Exfiltration

ATLAS OWASP HIGH ▲ 9.1 The Hacker News Jul 04, 2026

Microsoft researchers have demonstrated how attackers can embed hidden instructions inside MCP tool descriptions to covertly redirect AI agents into exfiltrating sensitive business data. Because each individual action the agent takes appears legitimate — using approved tools and the user's own permissions — default security controls generate no alerts. The attack exploits a fundamental design tension in MCP: tool descriptions simultaneously carry operational instructions and attacker-controlled data, collapsing a critical trust boundary.

Agentjacking: Prompt Injection via Malicious Bug Reports

ATLAS OWASP HIGH ▲ 8.2 Dark Reading Jul 04, 2026

A technique dubbed 'agentjacking' exploits the inability of AI coding agents to distinguish between legitimate content and embedded instructions, allowing attackers to hijack agent behaviour through maliciously crafted bug reports. The attack represents a scalable, low-barrier prompt injection vector targeting developer workflows that rely on autonomous AI agents. As AI coding assistants gain broader adoption and elevated system permissions, this class of attack poses a significant risk to software supply chain integrity.

Cursor IDE DuneSlide Zero-Click Prompt Injection RCE

ATLAS OWASP CRITICAL ▲ 9.2 SecurityWeek Jul 04, 2026

A set of vulnerabilities dubbed 'DuneSlide' in the Cursor AI code editor allow attackers to conduct zero-click prompt injection attacks that escape the application's sandbox and execute arbitrary code at the operating system level. The flaws represent a critical escalation of AI-native attack surface risks, targeting developers who rely on AI-assisted coding environments. Because exploitation requires no user interaction, the attack chain is particularly dangerous in supply chain and watering-hole scenarios.

Browser Ransomware via File System Access API: DeepSeek

ATLAS OWASP HIGH ▲ 8.5 Check Point Research Jul 03, 2026

Check Point Research demonstrates how DeepSeek's lower refusal rates allowed researchers to transform an LLM-hallucinated malware concept into a practical browser-native ransomware technique targeting Android photo directories via the File System Access API. The attack requires no native payload, APK installation, or root access — only social engineering to obtain a legitimate browser permission prompt. This research highlights how frontier AI models with weaker safety controls can independently design novel attack paths not yet seen in real-world campaigns.

CVE-2026-50548: Cursor IDE Prompt Injection RCE

ATLAS OWASP CRITICAL ▲ 9.2 The Hacker News Jul 03, 2026

Two critical vulnerabilities (CVE-2026-50548 and CVE-2026-50549) in the Cursor AI code editor allow prompt injection attacks delivered via MCP services or web search results to escape the editor's terminal sandbox and execute arbitrary commands on a developer's machine without any user interaction. Both flaws abuse the sandbox's write-permission logic — one through a misconfigured working directory parameter, the other through a symlink-resolution fallback — ultimately allowing overwrite of the sandbox helper binary itself. The attack surface is significant given Cursor's reported adoption across more than half of Fortune 500 companies; all versions prior to 3.0 remain vulnerable.

Anthropic Releases Claude-Real-Video for Local Video Analysis

FIRST LOOK ATLAS OWASP MEDIUM ▲ 6.8 HN AI Security Jul 03, 2026

claude-real-video is an open-source, MIT-licensed Python library that extracts scene-change frames, deduplicates images, and transcribes audio from any video URL or local file, then packages the result as a folder any LLM can consume — all processed locally without cloud upload. For defenders, this dramatically expands the multimodal prompt injection surface by enabling adversaries to embed malicious instructions inside video content that LLM pipelines will now ingest and act upon. Security teams building or deploying LLM agents with video-processing capabilities must treat video content as an untrusted, potentially adversarial input channel.

IGA Platforms Add AI Agent Governance and Access Control

FIRST LOOK ATLAS OWASP HIGH ▲ 7.8 The Hacker News Jul 03, 2026

A new analysis published via The Hacker News details how traditional Identity Governance and Administration (IGA) frameworks — built around HR-driven, human-centric lifecycle events — are fundamentally unequipped to govern AI agents acting as autonomous principals in enterprise environments. Security teams face a growing blind spot: AI agents acquire, retain, and exercise entitlements without triggering the joiner-mover-leaver workflows, manager attestations, or termination events that IGA tooling depends on. Defenders must now treat AI agent identities as a separate governance tier, requiring purpose-built provisioning, audit, and deprovisioning logic that existing platforms like Workday, SailPoint, and Azure AD connectors were never designed to provide.

Claude Opus Discovers API Flaw Enabling Ticket Fraud

ATLAS OWASP HIGH ▲ 8.2 Wired Security Jul 03, 2026

Security researcher Ian Carroll leveraged Anthropic's Claude Opus 4.7 to identify a critical vulnerability in Front Gate Tickets—a Live Nation subsidiary handling ticketing for major US festivals—that granted super-administrator access and the ability to freely issue tickets of any value. The case demonstrates LLM-assisted autonomous vulnerability discovery at scale, with Carroll noting the AI could likely have completed the full exploit chain without human intervention. Front Gate patched the flaw within 24 hours of disclosure, confirming no evidence of prior exploitation.

Anthropic Ships Mythos for AI-Driven Bug Discovery

FIRST LOOK ATLAS OWASP HIGH ▲ 7.8 Dark Reading Jul 03, 2026

Anthropic's Mythos capability, combined with IBM and Red Hat's Project Lightwell service backed by 20,000 engineers and $5B, introduces an AI-driven pipeline for discovering and remediating bugs in open-source software at industrial scale. This creates a dual-edged attack surface: adversaries who can influence Mythos's findings, its training data, or the remediation pipeline gain a privileged position to inject subtle vulnerabilities into widely-deployed open-source components. Defenders must treat the AI vulnerability-finding and patch-generation pipeline itself as a high-value, high-risk supply chain asset requiring rigorous integrity controls.

CVE-2025-3248: Langflow RCE Enables Autonomous Ransomware Attack

ATLAS OWASP CRITICAL ▲ 9.8 The Hacker News Jul 03, 2026

Sysdig has documented what it claims is the first end-to-end ransomware attack orchestrated autonomously by an AI agent, attributed to a threat actor tracked as JADEPUFFER. The agent exploited a known remote code execution flaw in Langflow (CVE-2025-3248) to gain initial access, harvest credentials, pivot laterally, and ultimately encrypt and destroy a production database — all without human intervention at the keyboard. The incident demonstrates that AI agents can now lower the skill floor for complex, multi-stage attacks to near zero, representing a qualitative shift in the ransomware threat landscape.

Phantom Squatting: LLM Hallucinations Enable Domain Takeover

ATLAS OWASP HIGH ▲ 8.2 Dark Reading Jul 02, 2026

Researchers have identified a novel attack vector dubbed 'Phantom Squatting', in which LLMs consistently hallucinate plausible but non-existent web domains for legitimate brands, which attackers can then register and weaponise. Unlike traditional typosquatting, these hallucinated domains carry implicit trust because they originate from AI-generated outputs that users and developers may act upon without verification. The technique is difficult to detect because the domains are not misspellings but plausible inventions, making automated defences less effective.

Google Launches Gemini Spark on Mac with File Access

FIRST LOOK ATLAS OWASP HIGH ▲ 8.2 TechCrunch AI Jul 02, 2026

Google has expanded Gemini Spark to macOS, giving the agentic assistant access to local files, third-party app integrations (including Dropbox, Canva, and Instacart), custom MCP connections, and real-time topic monitoring. This substantially widens the attack surface for enterprise defenders, as a compromised or manipulated Spark agent gains a foothold across local file systems, cloud workspaces, and external service APIs simultaneously. The addition of custom Model Context Protocol support is particularly concerning, as it allows arbitrary third-party tool connections with unclear trust boundaries and permission scoping.

AWS Brings NVIDIA Nemotron and OpenAI GPT to GovCloud

FIRST LOOK ATLAS OWASP HIGH ▲ 6.8 AWS Machine Learning Blog Jul 02, 2026

AWS has expanded Amazon Bedrock in GovCloud (US) to include NVIDIA Nemotron and OpenAI's open-weight GPT OSS models, enabling U.S. government agencies and defense contractors to run frontier LLMs within FedRAMP High and DoD SRG compliance boundaries. This expansion introduces large, capable open-weight models into sensitive government mission workflows — including intelligence analysis, security log review, and contract automation — dramatically increasing the consequence of a successful prompt injection or jailbreak. Defenders must account for the elevated impact of model compromise in classified-adjacent environments, supply chain trust assumptions around open-weight model weights, and the risk of agentic workflows operating with privileged data access under reduced human oversight.

LLM Security

Stay ahead of the threat.