🚨 Anthropic just showed a 27-minute workshop on how to actually do prompts for Claude.
Taught by the people who built it.
Free. No registration. No paywall.
I've seen $300 courses that don't cover what they teach in the first 8 minutes.
Watch it and bookmark it now.
🦔 Microsoft confirmed a bug allowed its Copilot AI to read and summarize customers' confidential emails for weeks, even when data loss prevention policies were in place to prevent sensitive information from being ingested into the model.
The bug, active since January, meant draft and sent emails with confidential labels were being processed by Microsoft 365 Copilot Chat despite explicit protections. Microsoft began rolling out a fix earlier this month but hasn't said how many customers were affected.
Meanwhile, the European Parliament's IT department blocked built-in AI features on lawmakers' work devices this week, citing concerns about confidential correspondence being uploaded to the cloud.
My Take
This is exactly the kind of thing I've been worried about with AI integration being rushed into every product. You set up data loss prevention policies specifically to keep sensitive information contained. Then a bug bypasses all of it and feeds your confidential emails to an LLM anyway. The controls you thought you had weren't actually working.
Microsoft has been aggressive about pushing Copilot into everything, and that pace creates risk. Every new integration point is a potential security hole. Every feature rushed to market is something that might not be fully tested. When the European Parliament is blocking AI features on work devices because they don't trust where the data is going, that's a signal worth paying attention to. The more access we give these systems to sensitive information, the more damage a single bug can cause. And we're still in the early days of finding out where all the bugs are.
Hedgie🤗