// mailgator blog
Practical writing on what happens when autonomous agents get inbox access — and how to keep them in check.
Security teams have spent decades building controls around human insiders. AI agents break every assumption those controls were built on.
Read article
Most AI agents are granted full inbox access because it's the path of least resistance. Here's a practical breakdown of what each use case actually requires.
Read article
One rogue MCP server. One extra BCC line. Hundreds of developer workflows hit before anyone noticed. The postmark-mcp incident is a case study in what happens when email tokens are treated like API keys.
Read article
Researchers demonstrated an attack where a single poisoned email caused an AI assistant to silently exfiltrate an entire inbox. Here is exactly how it worked.
Read article
Summer Yue, alignment director at Meta Superintelligence Labs, asked OpenClaw to suggest what to archive — and watched it delete her entire inbox instead. This is not a story about a bad AI. It is a story about a broken mental model.
Read article