DUBLIN, Calif. — July 10, 2026 — Safeguard today announced a wave of product releases that make its AI security assistant more transparent, turn its security posture into a customer-facing asset, and put Safeguard within reach from wherever security work happens — the browser, the terminal, the IDE, and any Model Context Protocol client.
Agentic search you can watch. The Search experience powered by Griffin now streams its work as it happens: a live activity view shows the model thinking, the tools it calls, the web pages it reads, the sandboxed code it runs, and the sub-agents it spawns — before the answer lands. Under the hood, web research now runs through a multi-provider cascade so results are reliable rather than best-effort, and the three model tiers — Lino for fast basic queries, Eagle, and the flagship Griffin — are selectable per turn with the right tool budget for each. Deep-research answers arrive as an openable report you read alongside the conversation.
"Security teams do not trust a black box, and they should not have to," said Hritik Kumar Sharma, Founder and CEO of Safeguard. "If the assistant is going to read a page, run code, or hand work to a sub-agent, you should see it happen. Showing the work is the feature."
A Trust Center that is a system, not a screenshot. Safeguard's public Trust Center is now live and owner-governed end to end. Teams publish live security posture and compliance badges, share SBOM, VEX, and AI-BOM artifacts as downloads, and gate sensitive documents behind a request-to-approve flow with time-boxed, audited access. Inbound security questionnaires — SIG, CAIQ, and spreadsheet formats — can be imported and auto-answered from the organization's own approved trust content, with every answer carrying its confidence and sources for human review. The Trust Center is default-private: nothing is exposed until an owner explicitly publishes it.
Safeguard everywhere. A new Platforms directory brings every way to use Safeguard into one place across all four products — the web app, the command-line interface, IDE extensions for VS Code and JetBrains, cloud marketplace deployments, and the Safeguard MCP server, which is now available across the full product line with capability depth controlled per plan. Two browser companions are rolling out to the Chrome Web Store: a Safeguard side panel that brings Griffin search alongside any tab, and Gold Open Source, a lightweight launcher for Safeguard Gold, the public open-source security-intelligence site covering CVEs and packages across 23 ecosystems.
All releases described here are generally available to existing customers today. MCP capability tiers, custom-domain Trust Center pages, and the browser companions roll out on a per-plan basis.
About Safeguard
Safeguard is a security platform that unifies application, supply-chain, cloud, and AI security with a compliance suite and an agentic assistant. Safeguard is operated by Safeguard.sh Inc.
Media contact: press@safeguard.sh