Zero Trust

Weekly roundup covering enterprise AI guardrails and agent security, sovereign AI platforms, security advisories, and AI infrastructure market signals.

Zero trust is presented as an operational model, with user, identity, and data sprawl cited as barriers to implementation.

The blog argues that zero trust for AI agents should cover identity, context-based access, inline inspection, and runtime guardrails.

Gigamon and Zscaler integrated ZPA with Gigamon AMI to add application activity visibility for Zero Trust and hybrid cloud environments.

Netskope says zero trust models for human access don’t cover AI agents using MCP and app-to-LLM traffic, and describes related visibility controls.

Zscaler expanded its Zero Trust SASE with the ZAgent Framework and added browser, B2B exchange, endpoint sandbox, and multi-cloud workload features.

Cequence Security said AI security focus should shift from authentication to controlling AI agent behavior using its AI Gateway approach.

Zscaler said it extended its Zero Trust Exchange to secure agentic AI with Zscaler AI Broker, Endpoint AI Security and an AI Access Graph.

CISA reinforced that zero trust must be adapted for OT, including legacy and segmented networks, and outlined five principles for federal agencies.

Netskope says Threat Labs data shows AI apps grew fivefold, while it uses Claude Mythos Preview in Glasswing to identify and fix Netskope vulnerabilities.

Xage Security announced support for NVIDIA DOCA security and NVIDIA Vera BlueField-4 STX to provide line-rate visibility and policy enforcement for agentic AI. The integration is described across Xage components and DOCA services, aiming for identity-based Zero Trust across AI factory environments without host software changes.

Xage Security updated its Zero Trust for AI platform with Agent Sentry and Resource Gateway for agent visibility and control.