Nvidia has launched an Open Agent Safety Platform designed to constrain autonomous systems even when application-level safeguards fail. The company says the platform combines OpenShell, an open-source runtime that traces actions and enforces policy, with Sentry, a reference design that uses BlueField-4 data-processing units as an independent watchdog.

The architectural point matters more than the product branding. OpenShell is intended to isolate agents and control their access to models, tools and data. Sentry sits outside the agent and host software, correlating activity and quarantining agents that cross defined boundaries. Nvidia says OpenShell can be extended to Arm and Intel environments, while Sentry is tied to its BlueField hardware. Those claims will need independent testing, particularly around policy bypasses, latency and operational failure modes.

The launch also creates an unusually broad enterprise coalition. Nvidia named Anthropic, Cisco, CrowdStrike, HPE, JPMorganChase, Microsoft, Palantir, Palo Alto Networks, Salesforce and ServiceNow among participating organisations. For regulated firms, the important question is whether out-of-band enforcement can provide stronger evidence of control than prompt-layer guardrails alone. The platform is therefore best read as a proposed control plane for agentic systems, not as proof that autonomous deployment is now safe.

Execution Level Governance- What audit-ready agent governance actually looks like
David Girvin, founder and CEO of Assury argues that model-in-the-loop review, AI governing AI, is fundamentally unreliable for regulated environments: even the best-performing models miss a meaningful share of violations, the reviewing model is typically provided by the same vendor being reviewed, and prompt injection or context poisoning can compromise both the acting agent and its supposed overseer simultaneously. He makes the case for deterministic, architecturally enforced controls instead, walking through Assury’s approach of autonomy zones, session risk accumulation, and credential starvation, which lets a compromised agent be cut off from its tools instantly rather than relying on time-boxed access. The conversation touches on why David is sceptical of just-in-time credentialing as a solution for agent security more broadly, since agent sessions don’t run on predictable human timescales, along with the current gap between how identity and security vendors are pitching agent protection and what he sees happening at the execution layer in practice. He also discusses the compliance and audit implications of probabilistic decision-making, arguing that regulated industries will increasingly need tamper-evident, hash-chained audit trails that can withstand scrutiny from auditors and regulators who are only beginning to understand agentic risk, and reflects on a named frontier lab’s own published framework as an example of the gap between research and practitioner reality. Elsewhere, David reflects candidly on building a bootstrapped security company in an increasingly crowded market, why he turned down aggressive VC funding to stay in control of the product, and what a credible third-party assessment of his own gateway would need to look like given that Assury sits directly in the execution path for every customer’s agents.
BrightTALK
Share this post
The link has been copied!