When we chat with an AI, it's easy to treat it like a search engine that talks back. We ask a question, it spits out an answer with absolute authority, and we move on.
But there's a hidden danger in that experience: AI models naturally default to the "Voice of God." They deliver facts, theories, and unverified data in the exact same confident tone, leaving out crucial context, source attribution, or alternative viewpoints.
If we want to trust the information we get from AI, we have to teach it media literacy. By injecting a system-level prompt into your chat settings, you can force the AI to cite its sources, communicate its level of uncertainty, and present balanced perspectives on complex topics.
Here is how a simple prompt changes the game, and exactly how to add it to ChatGPT, Claude, and Gemini.
By establishing these rules at the system level, you fundamentally change how the AI constructs its responses. It acts as a cognitive brake, forcing the AI to follow three core pillars:
"AI is an incredibly powerful research partner, but only if you hold it to the same standards as a rigorous journalist."
You don't need to copy and paste these rules into every single prompt. You can bake them directly into the background of your favorite chats.
OpenAI gives you a direct way to control the AI's default behavior across all new chats.
Where to go: Click your profile icon → Settings → Custom Instructions (or Personalization).
How to paste it: You will see two boxes. Leave the first one ("What would you like ChatGPT to know about you?") as it is. Paste the media literacy prompt directly into the second box: "How would you like ChatGPT to respond?"
Note: If you use the premium version, you can also create a custom "GPT" in the Explore sidebar and paste this into the "Instructions" section.
Anthropic allows premium users to build dedicated workspaces called "Projects" that follow strict background instructions.
Where to go: In the left sidebar, click Projects → Create Project.
How to paste it: Once your project is created, look at the right-hand panel for Set Project Instructions. Paste the entire system prompt there. Every chat you start inside that specific project will automatically inherit those media literacy guardrails.
Google handles persistent behavioral instructions through a feature called "Gems," allowing you to build specialized versions of the AI.
Where to go: In the left sidebar, scroll down and click Gems Manager (or Create a Gem).
How to paste it: Name your Gem something like "Literacy Bot" or "Research Assistant." In the Instructions box, paste the system prompt. Whenever you need to research a factual or nuanced topic, just open this specific Gem.
AI is an incredibly powerful research partner, but only if you hold it to the same standards as a rigorous journalist. Giving your chat a permanent media literacy prompt ensures you get clear, sourced, and honest context every single time you hit enter.