AI Security
How do you stop forgotten scheduled AI tasks from running forever?
Some assistants expire or switch off idle schedules, others do not. Build an owner, a review date and a kill path for every recurring AI task.
Oct 19, 20266 min read
Resources
Supply Chain Security, in plain English.
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.
Filter
Articles
RSS feedAI Security
A one-week bake-off plan for AI assistant agent features
A day by day one-week plan to test scheduled tasks, file generation, the desktop app and remote sandboxes in AI assistants before you commit.
Oct 19, 20266 min read
AI Security
What state survives when a Claude Code cloud session goes idle?
Idle Claude Code cloud sessions pause, then expire. What is restored, what is lost, and what that means for secrets, caches, and long running agent tasks.
Oct 19, 20266 min read
AI Security
Is an AI spend cap a security control for leaked API keys?
A spend cap and rate limits bound the damage from a leaked AI API key or a runaway agent. They are a backstop, not a substitute for key hygiene.
Oct 19, 20268 min read
AI Security
Can an AI assistant create Office files with macros? What to ask
Macros in files from the internet are blocked by default in Microsoft 365 apps. Here is what to ask a vendor whose AI assistant creates Office documents.
Oct 19, 20266 min read
AI Security
Can a coding agent steal cloud credentials from the metadata service?
If an agent sandbox runs on a cloud VM, the instance metadata service may hand out credentials. How IMDSv2 and a hop limit narrow that path.
Oct 19, 20267 min read
AI Security
What leaves your machine when you send a repo to a cloud agent?
Starting a cloud agent task can upload a clone or a bundle of your local repository, history and edits included. What to check before you press go.
Oct 19, 20266 min read
AI Security
Why do security connector calls time out in chat?
Big scans and long finding lists can be slow, and servers must rate limit while clients set timeouts. What to expect, how to test and what to ask.
Oct 19, 20266 min read
AI Security
Can an AI chat connector replace a security dashboard?
No. A connector is a fast way to ask questions of security data, not a replacement for views, audit trails and approvals. Where chat helps and where not.
Oct 19, 20266 min read
Stay informed
Weekly insights on software supply chain security, delivered to your inbox.
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.