Power Up Your Protection - Level 3: Collaboration in an Agentic Workspace
Quick Overview
- Covers AI browsers and assistants, AI jailbreaks, and how chatbots handle your data
- Opens Monday, Oct. 19 in ZenGuide; start it from your Level 3 email from IT Security
- Takes about 10 minutes, plus a short newsletter attached to this article
- Optional; all levels should be completed by Friday, Oct. 30
AI tools are moving from answering questions to taking action: browsing the web, summarizing documents, and working across our apps. That convenience comes with new risks.
In This Level
- 60 Seconds to Better Security: AI Browser Risks (video)
- AI Security Minute: Universal vs. Non-Universal Jailbreaks (video)
- Generative vs. Retrieval Chatbots (interactive module)
- Bonus: Using AI Chatbots Safely newsletter, attached to this article
Know the Signs
- An AI assistant that can read your screen, email, or files can be tricked by hidden instructions in a web page or document
- AI browser extensions often ask for access to every site you visit
- Attackers write “jailbreak” prompts that make AI tools ignore their own safety rules
- Generative chatbots can make things up. Retrieval chatbots can repeat anything in the documents they can reach.
Protect Yourself
- Review what an AI assistant is about to do before you approve it, especially sending messages, sharing files, or making purchases
- Before you use a chatbot, ask: what data can it reach, and where do my prompts go?
- Remove AI extensions you don’t use
|
!
|
Important
Only connect AI tools to your SFA email, files, or accounts if the tool is approved for that use. See SFA’s AI Use Guidelines. |
This Week’s Challenge
- Audit your extensions: remove browser extensions on your work computer that you don’t use or recognize.
- Check an AI connection: for one AI tool you use, turn off access to anything it doesn’t need.
- Read the newsletter and try its activity.
|
i
|
Tip
Let AI help, but keep your hands on the wheel.
|
Need Help?