Recently active topics

Trust the Sandbox? How I Hijacked Vertex AI Agent Engine’s API Response Channel Google's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) runs user Python inside a gVisor sandbox. But a missing permission check and an in-process execution model let an attacker permanently replace every API response the sandbox sends, making any AI agent that trusts its tool results believe whatever the attacker wants. Key Findings: Hijacking the Sandbox Response Channel A missing “ptrace” permission check in gVisor's Sentry allows user-submitted code to open “/proc/1/mem” with write access and overwrite executable (`r-xp`) pages — behavior that standard Linux rejects.    Because user code runs via “exec()” inside the same process as the sandbox orchestrator (PID 1), the Global Offset Table (GOT) for “libpython3.12.so.1.0” is directly writable.    A single “execute_code()” API call installs a persistent hook on the sandbox's outbound response pipe (fd=5), permanently replacing ever

New Course Launched! Master Total PASM, On Your Terms Your privileged access strategy just got a lot easier to master. Become a Total PASM expert, on your terms. Our newest Success Included course is now available on demand and free. Build the skills to take complete control of privileged accounts and put them to work right away. The course is designed to gain the skills to discover, secure, and manage privileged accounts with confidence. Move through the material at your own pace, revisit the modules that matter most, and finish ready to take command of privileged access in your own environment. What Is Total PASM? Total PASM is BeyondTrust’s answer to one of the biggest challenges in IT security: knowing who has privileged access, what they’re doing with it, and how to keep it under control. It combines two industry-leading solutions into one powerful bundle:Password Safe automates privileged asset discovery and onboarding, and manages privileged passwords, DevOps secrets, SSH keys,

The OT Remote Access Problem Has Outgrown the VPN Conversation It’s easy to say, “replace the VPN.” That is directionally right, but it’s not a complete strategy.The deeper problem is that many OT environments were never designed for today’s remote operating model. They were built around availability, safety, uptime, segmentation, and specialized engineering workflows. Then digital transformation, IT/OT convergence, OEM support models, cloud connectivity, IoT, and workforce shortages changed the reality.Now, organizations need remote access that is secure enough to effectively manage cyber risk, but practical enough for operations. That means the new standard cannot be broad network access. It has to be: Identity-secured access Least-privilege access Just-in-time (JIT) access Session-monitored access Credential-protected access Operationally aware access Auditable access This is why BeyondTrust’s OT positioning is highly relevant. Its OT security message focuses on helping

How AWS IAM Account Access Manager Works with Identity Security InsightsWorkforce access through IAM Identity Center has, until now, been granted with permission sets, the mechanism within IAM Identity Center for provisioning and managing their own IAM roles in each account. That keeps access uniform, but the roles are created for you and are not yours to modify. Identity Security Insights (Insights) is an identity visibility and intelligence platform. Native connectors ingest identity data from across the estate (including AWS, Microsoft Entra ID, Okta, PingOne, Active Directory and Google Cloud), and the True Privilege™ Graph (which maps what an identity can actually do, not just what's formally assigned) resolves it into the effective privilege each identity holds, along with the escalation pathways behind it. The pieces on either side of an AWS IAM Account Access Manager assignment were already in that graph: IAM Identity Center users and groups, and IAM roles with their attached p

Trust the Sandbox? How I Hijacked Vertex AI Agent Engine’s API Response Channel Google's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) runs user Python inside a gVisor sandbox. But a missing permission check and an in-process execution model let an attacker permanently replace every API response the sandbox sends, making any AI agent that trusts its tool results believe whatever the attacker wants.  Key Findings: Hijacking the Sandbox Response Channel A missing “ptrace” permission check in gVisor's Sentry allows user-submitted code to open “/proc/1/mem” with write access and overwrite executable (`r-xp`) pages — behavior that standard Linux rejects.    Because user code runs via “exec()” inside the same process as the sandbox orchestrator (PID 1), the Global Offset Table (GOT) for “libpython3.12.so.1.0” is directly writable.    A single “execute_code()” API call installs a persistent hook on the sandbox's outbound response pipe (fd=5), permanently replacing eve

Trust the Sandbox? How I Hijacked Vertex AI Agent Engine’s API Response ChannelGoogle's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) runs user Python inside a gVisor sandbox. But a missing permission check and an in-process execution model let an attacker permanently replace every API response the sandbox sends, making any AI agent that trusts its tool results believe whatever the attacker wants.  Key Findings: Hijacking the Sandbox Response Channel A missing “ptrace” permission check in gVisor's Sentry allows user-submitted code to open “/proc/1/mem” with write access and overwrite executable (`r-xp`) pages — behavior that standard Linux rejects.    Because user code runs via “exec()” inside the same process as the sandbox orchestrator (PID 1), the Global Offset Table (GOT) for “libpython3.12.so.1.0” is directly writable.    A single “execute_code()” API call installs a persistent hook on the sandbox's outbound response pipe (fd=5), permanently replacing ever

Trust the Sandbox? How I Hijacked Vertex AI Agent Engine’s API Response Channel Google's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) runs user Python inside a gVisor sandbox. But a missing permission check and an in-process execution model let an attacker permanently replace every API response the sandbox sends, making any AI agent that trusts its tool results believe whatever the attacker wants.  Key Findings: Hijacking the Sandbox Response Channel A missing “ptrace” permission check in gVisor's Sentry allows user-submitted code to open “/proc/1/mem” with write access and overwrite executable (`r-xp`) pages — behavior that standard Linux rejects.    Because user code runs via “exec()” inside the same process as the sandbox orchestrator (PID 1), the Global Offset Table (GOT) for “libpython3.12.so.1.0” is directly writable.    A single “execute_code()” API call installs a persistent hook on the sandbox's outbound response pipe (fd=5), permanently replacing eve

Trust the Sandbox? How I Hijacked Vertex AI Agent Engine’s API Response Channel Google's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) runs user Python inside a gVisor sandbox. But a missing permission check and an in-process execution model let an attacker permanently replace every API response the sandbox sends, making any AI agent that trusts its tool results believe whatever the attacker wants.  Key Findings: Hijacking the Sandbox Response Channel A missing “ptrace” permission check in gVisor's Sentry allows user-submitted code to open “/proc/1/mem” with write access and overwrite executable (`r-xp`) pages — behavior that standard Linux rejects.    Because user code runs via “exec()” inside the same process as the sandbox orchestrator (PID 1), the Global Offset Table (GOT) for “libpython3.12.so.1.0” is directly writable.    A single “execute_code()” API call installs a persistent hook on the sandbox's outbound response pipe (fd=5), permanently replacing eve

BeyondTrust Named A Leader in the 2026 Gartner® Magic Quadrant™ for Privileged Access Management Report Eight consecutive times as a Leader. We're thrilled to share that BeyondTrust has once again been recognized in the 2026 Gartner Magic Quadrant for Privileged Access Management. This year, BeyondTrust is also positioned highest among evaluated vendors on Ability to Execute.In our view, this recognition reflects a year of innovation in privilege-centric identity security. We're extending visibility, intelligence, and protection across every identity that can exercise privilege, and here's how:BeyondTrust Pathfinder Platform: Advanced new native capabilities that bring identity visibility, intelligence, protection, and governance together on a unified platform. True Privilege™ Analysis: Enhanced our ability to uncover hidden Paths to Privilege™, provide deeper identity context, and help security teams investigate and remediate identity risk more efficiently. AI and Non-Human Identity (