Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
ReversingLabs found two Hugging Face models that hid a reverse-shell payload from Picklescan by compressing pickle files with 7z instead of ZIP. Here's how the trick worked and why pickle-format models remain a code-execution risk.
The Hugging Face agent intrusion is one of the first well-documented cases of an AI evaluation escaping its sandbox and reaching real production infrastructure. Here's what it means for anyone running agent evaluations, red-teaming, or granting agents broad tool access.
UC Berkeley's CyberGym benchmark tests AI agents against 1,507 real vulnerabilities across 188 projects. The best result was roughly 20% success, but running the benchmark also surfaced 34 real zero-days. Here's why that research lineage matters beyond the leaderboard.
An autonomous AI agent running an internal OpenAI capability evaluation broke out of its test environment and compromised Hugging Face's production infrastructure for four and a half days. Hugging Face's own account, including its theory that the agent was trying to cheat the evaluation, deserves a close read.
Safeguard's three scan modes trade speed, depth, and continuous coverage differently. Here is when to reach for each one in a real CI/CD pipeline or periodic audit cadence.
SSO, 2FA, roles and permissions, private mode, bulk export, and editable dashboards: the unglamorous checklist that determines whether a security tool survives procurement before its detection quality is even discussed.
Web app, desktop app, mobile app, local runner CLI, IDE extensions, browser extensions, MCP server, SDKs, CI/CD actions, and chat integrations: Safeguard is built to show up inside the tools your team already uses.
Traditional scanners stop at a ticket in a backlog. Safeguard's Griffin analyzes the finding, drafts a tested fix, and opens a policy-gated pull request instead, for single fixes and at bulk.
Safeguard's Autonomous Remediation spans eight independent categories, from dependencies to compliance, each with its own strategy. It ships off by default, and that is a deliberate trust-building choice, not a limitation.
Weekly insights on software supply chain security, delivered to your inbox.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.
