All news with #agent security tag

🔒 Enterprises are confronting a shift: autonomous AI agents now operate inside corporate environments with real permissions and real consequences. Security must move beyond build-time controls to continuous runtime monitoring that observes agent behavior, preserves tamper-proof logs, and applies agent-aware policies. Practical first steps include inventorying agents, extending EDR-style behavioral baselining, and designing incident-response playbooks that stop misbehaving agents without destroying evidence.

🛡️Security validation must evolve from disconnected tests to continuous, context-aware assessment powered by agentic AI. The piece argues that defenders need to converge three perspectives — adversarial, defensive, and risk — into a unified discipline supported by a Security Data Fabric that unites Asset Intelligence, Exposure Intelligence, and Security Control Effectiveness. With real-time context, autonomous agents can plan, execute, and prioritize validation workflows, turning fragmented tool outputs into actionable evidence and faster remediation. The article highlights Picus Security and industry recognition as indicators that the market is moving toward CTEM-native, agentic validation.

🔒 China's CNCERT warned that OpenClaw, an open-source, self-hosted autonomous AI agent, ships with weak default security and broad system privileges that attackers can abuse to seize endpoints and exfiltrate data. The advisory highlights indirect prompt injection (IDPI/XPIA) risks where benign features like web-page summarization and messaging link previews are weaponized to embed malicious instructions or automatically leak secrets. Researchers at PromptArmor demonstrated a technique in which an agent constructs attacker-controlled URLs that, when rendered as link previews, transmit confidential data without user clicks. CNCERT also flagged risks from malicious skills, accidental destructive commands, and disclosed vulnerabilities, urging isolation, tightened network controls, credential protection, and cautious skill sourcing.

🛡️ Organizations are adopting agentic AI—autonomous, task-driven systems powered by LLMs—to streamline processes and boost throughput. These agents can plan, act, and iterate, but their non-deterministic behavior creates gaps in traceability, auditability, and access control. Apply strong role-based access, threat modeling, and oversight (human or independent evaluators) to limit exposure and ensure safe deployment.

🔒 AI agents are transforming workflows but can act as an unmonitored access layer—an 'invisible employee' with broad privileges. In an upcoming webinar, Rahul Parwani, Head of Product for AI Security at Airia, will explain how attackers are manipulating agents to exfiltrate sensitive information and how to stop them. Attendees will learn the Dark Matter of identity, common manipulation techniques, and a practical safety blueprint to limit privileges, detect misuse, and prevent leaks. Reserve your spot to learn actionable defenses.

🎧 Microsoft introduces The Shift, an evolution of its earlier podcast to explore agentic AI across engineering, product, and strategy perspectives. Over eight weekly episodes this spring, hosts and guests from teams including Microsoft Azure, Microsoft Fabric, and Microsoft Foundry tackle practical questions about how agents interact with data, databases, and cloud foundations. Episodes emphasize that agents succeed only when data strategy, cloud reliability, and application orchestration work together, highlighting operational concerns like observability, governance, security, and optimization.

🤖 Amazon Lightsail now lets you deploy OpenClaw, a private self-hosted AI assistant, on your own cloud infrastructure with simple, secure defaults. Each Lightsail OpenClaw instance includes built-in security controls—sandboxed agent sessions, one-click HTTPS for TLS, device-pairing authentication, and automatic snapshots—reducing manual configuration and operational risk. Amazon Bedrock is the default model provider, and users can swap models or connect to Slack, Telegram, WhatsApp, and Discord as needed.

🧭 This practical guide from GoogleWalks developers through how to move AI agents from prototype to production, highlighting architecture, operational patterns, and safety considerations. It explains an agent as an LLM-driven autonomous system surrounded by an orchestration layer that manages session state, long-term memory, retrieval (RAG), tool use, and security. The post emphasizes emerging interoperability standards such as MCP and A2A, and underscores the importance of context engineering, trajectory-based testing, and staged rollouts. Authors provide targeted guides and code samples to help teams adopt these practices and validate agents before broad deployment.

🔐 Today’s enterprise AI agents are no longer passive assistants but active operators that authenticate to systems using API keys, OAuth tokens, cloud roles, and service accounts. The article advocates treating every agent as a first-class identity with unique identities, lifecycle management, defined roles, clear ownership, and auditability. It warns that identity alone is insufficient because agents are dynamic and can drift from original missions; instead it promotes intent-based permissioning, activating privileges only when an agent's declared mission and runtime context justify the action. Practical steps include inventorying agents, assigning lifecycle-managed identities, documenting approved missions, and enforcing conditional access based on identity, intent, and context.

🔒 Self-hosted agent runtimes such as OpenClaw shift the execution boundary by ingesting untrusted text, downloading third‑party skills, and acting with the host's credentials. This combination makes the runtime effectively untrusted code execution with persistent tokens and elevated access, unsuitable for standard workstations. Microsoft recommends evaluating OpenClaw only in isolated VMs or dedicated devices, using dedicated non‑privileged credentials, continuous monitoring, and a fast rebuild plan. Prioritize containment, least privilege, and monitoring with solutions like Microsoft Defender XDR.

🔒OpenClaw has patched six vulnerabilities disclosed by Endor Labs, including SSRF, missing webhook authentication and a path traversal issue that range from moderate to high severity. The set includes CVE-2026-26322 (Gateway SSRF, CVSS 7.6), CVE-2026-26319 (Telnyx webhook auth bypass, CVSS 7.5) and several GitHub Security Advisories such as GHSA-56f2-hvwg-5743. Endor warns that agent frameworks’ multi-layered architectures mean vulnerabilities can span files and components, requiring data-flow analysis and layered validation to mitigate exploitation. SecurityScorecard also flagged many publicly exposed OpenClaw instances, raising enterprise risk.

🚀 Amazon Bedrock now offers Mistral AI Devstral 2 123B, an open-weight 123B-parameter LLM optimized for agentic software engineering workflows. The model focuses on code generation, automation, and reliable multi-step reasoning, supporting long-context comprehension for multi-turn coding tasks. Bedrock exposes Devstral 2 via a single, fully managed API so customers do not need to provision infrastructure or host models. It is intended for production coding assistants, automated code review, and complex software development agents and is available in select AWS Regions.

🔒 Traditional endpoint defenses miss a growing class of non-binary software — browser extensions, code packages, IDE plugins, local servers, containers and model artifacts — that employees and developers install without centralized oversight. AI agents amplify that blind spot by acting with user credentials, autonomously discovering, invoking and installing components at machine speed. Palo Alto Networks says it intends to acquire Koi to deliver Agentic Endpoint Security, focused on visibility, continuous risk analysis and real-time policy enforcement to remediate risky behaviors.

🛡️ Late 2025 marked a decisive shift from brittle RAG deployments to autonomous, goal-oriented agents across the enterprise. While architectures like self-RAG and CRAG improved reliability, they also expanded the attack surface to include every document, memory store and integrated tool. New threats — indirect prompt injection, memory poisoning and agentic DoS — can exfiltrate data or drain budgets, forcing defenders to secure the full perception-reason-action loop.

⚠️ OpenClaw is an open‑source AI‑agent orchestration tool that runs locally, integrates with common chat apps and can use any LLM backend, driving rapid adoption. Researchers have found widespread exposed instances, critical authentication‑bypass flaws, plaintext credentials in the ClawHub marketplace and hundreds of malicious skills enabling credential theft and remote code execution. Experts urge enterprises to ban or tightly restrict use, enforce least privilege, MFA, endpoint segmentation and continuous telemetry if pilots are allowed.

🔓 Hudson Rock says an info‑stealer, likely a Vidar variant, exfiltrated an OpenClaw agent's configuration, including openclaw.json, device.json and soul.md. The files contain gateway tokens, cryptographic keys and the agent's operational 'soul,' which could let attackers impersonate the AI assistant or connect to local instances if exposed. The incident signals a shift from stealing credentials to harvesting AI agent identities, and vendors should expect targeted modules to follow.

🔐 Hudson Rock has observed information-stealing malware exfiltrating configuration and memory files from the OpenClaw agent framework, exposing API tokens, private keys, and persistent agent memory. The activity, attributed to a Vidar-like infostealer and recorded on 13 February 2026, captured openclaw.json, device.json, and agent 'soul' and memory files. With these items an attacker could impersonate the device, bypass Safe Device checks, access encrypted logs, or fully compromise a user's digital identity. Organizations should audit agent directories, apply vendor fixes, and enforce strict filesystem permissions immediately.

⚠️ OpenClaw (formerly Clawdbot/Moltbot) is an open-source local AI assistant that integrates with chat apps and can access calendars, email, browsers and the filesystem. Since its November 2025 debut and January 2026 viral spike, multiple critical vulnerabilities — notably CVE-2026-25253 — enabled token theft and arbitrary command execution. The project stores secrets in plaintext, exposes dangerous defaults, and hosts a marketplace where malicious skills have proliferated. Organizations face regulatory, operational, and insider-threat risks if employees run this software on personal or corporate devices.

🔒 The Microsoft Defender Security Research Team describes the top 10 misconfigurations that make Copilot Studio agents risky across enterprises. The post explains how small choices — broad sharing, weak authentication, raw HTTP calls, hard-coded secrets, orphaned agents, and unconstrained orchestration — create exploitable paths. It includes Advanced Hunting Community Queries to detect these issues and a short mitigation checklist to reduce exposure. The guidance stresses treating agents as production assets with lifecycle governance and least-privilege controls.

⚠️ TrendAI warns that so-called AI skills—executable artifacts that combine human-readable instructions, decision logic and operational constraints—are dangerously exposed to theft, sabotage and disruption. These skills power automation in tools such as Anthropic’s Agent Skills, OpenAI’s GPT Actions and Microsoft’s Copilot Plugin, and can surface proprietary data and business logic. If attackers obtain skill logic or operational data they could disrupt public services, manipulate manufacturing or steal sensitive records. TrendAI recommends integrity monitoring, strict access controls, separation of data and logic, least-privilege execution, adversary testing and continuous logging and auditing.