# Habit 0: Protect the Commons → Reflect Before You Reveal #### What It Means in LLM Context Before practicing any of the 7 habits, ask yourself: **“Should I even share this?”** Not all LLMs are created equal, and **not all platforms protect your prompts equally**. Some allow toggling off training.\ Some claim zero retention.\ Some anonymize by design.\ Others don’t. But **none of them are immune** to misuse, misconfiguration, or subpoena. So before you begin your thoughtful prompting journey, sharpen this first instinct: “Am I about to feed something I can’t unfeed?” *** #### Common “Over-Share” Scenarios | Prompt Style | Hidden Risk | | ---------------------------------------- | ------------------------------------------------------------------------ | | “Here’s a Slack thread from my team” | Leaks names, company strategy | | “This is a draft of my lawsuit response” | Reveals legal positioning | | “Help me rewrite my performance review” | Discloses workplace identities, evaluation tone, and internal benchmarks | | “Fix this patient note” | Violates HIPAA or medical confidentiality | | “Here’s my dating profile + chat log” | Combines PII, behavioral data, and emotional metadata | Most users don’t realize what they’re revealing **until it’s too late.** *** #### Safety Habits
HabitExample
Mask as default“Here’s a simulated conflict between two coworkers.”
Check your pasteBefore sending, scan: “Any names, IDs, or direct quotes?”
Partial disclosureInstead of 10 paragraphs, try 1 representative chunk
Swap context type“This is like a project kickoff” instead of “My team’s actual meeting”
Declare simulationStart with: “Let’s treat this as a fictional example…”
These moves preserve fidelity **without inviting leakage.** *** #### Understand Privacy Isn’t Binary Not all platforms are equal, and **privacy toggles ≠ leakproof systems.** Examples: * Some models respect “don’t train on this” toggles, others may store for safety/abuse reasons * Some vendors and tools let you disable training + history, always check with the user privacy section carefully. * API-based access typically has stricter data handling, but still passes through cloud infra * Enterprise contracts offer better protection, but may still log metadata **Assume partial visibility unless you control the full stack.** *** #### Self-Check Prompts * “Would I be okay with this prompt being screenshot and shared publicly?” * “Does this data belong to me, or to someone else?” * “If this prompt showed up in a future model’s output, would that be okay?” *** #### Integration Cue *Before Habit 1, even before you type, ask:*\ \&#xNAN;*“Would I be okay if this prompt, or its contents, showed up in another model, product, or output later on?”* The commons you’re protecting is not just **your privacy**, it’s everyone’s model safety. *** --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://praxistutor.gitbook.io/llm/reframing-the-7-habits-for-reflective-llm-use/section-2-the-seven-habits-of-reflective-llm-use/habit-0-protect-the-commons-reflect-before-you-reveal.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.