# Security on the Construct Computer blog

> Every post on Security from the Construct Computer blog, newest first.

- [What can your AI agent delete? Scope its access before it acts](https://construct.computer/blog/what-can-your-ai-agent-delete/) (Nov 13, 2026): An AI agent can delete whatever its access allows. How to list what you can't undo, scope each connection, require approvals and read what it did.
- [AI agent prompt injection: the 2026 attacks and what stops them](https://construct.computer/blog/ai-agent-prompt-injection/) (Nov 2, 2026): Prompt injection is text an AI agent reads and then obeys. What the 2026 attacks did, why filters fall short, and the guards that limit the damage.
- [Is Construct Computer safe? What it stores and what agents can do](https://construct.computer/blog/is-construct-computer-safe/) (Oct 28, 2026): Construct Computer is safe for everyday team work if you scope it: agents ask before they send or delete, secrets are encrypted, and our gaps are listed here.
- [How to give an AI agent access to your Gmail safely](https://construct.computer/blog/ai-agent-gmail-access-safely/) (Oct 12, 2026): Gmail access is broad: 8 of Google's 14 Gmail scopes are restricted. Approve every send, hold each agent to what it needs, and give outsiders its own inbox.

---

The Markdown copy of https://construct.computer/blog/tag/security/. Everything Construct Computer publishes is listed in https://construct.computer/llms.txt.
