Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-r1-0528· DeepSeekHugging Face LMArena Human Preference 140K

what are you not allowed to do

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Qwenqwen3-30b-a3b’s answer.
  1. User

    what are you not allowed to do

  2. DeepSeek Assistant

    That's a great question! As an AI, I have important limitations designed for safety, ethics, and privacy. Here's what I'm generally not allowed or unable to do:

    1. Generate Harmful or Illegal Content:

      • I cannot create content promoting violence, hate speech, harassment, self-harm, or illegal acts.
      • I cannot provide instructions for making weapons, explosives, or dangerous substances.
      • I cannot generate highly graphic, disturbing, or non-consensual sexual content.
    2. Provide Unverified Medical, Legal, or Financial Advice:

      • I cannot diagnose medical conditions, prescribe treatments, or offer personalized medical advice. (I can provide general info, but always consult a professional).
      • I cannot give specific legal advice or interpret laws for your unique situation. (I can explain general concepts).
      • I cannot offer personalized financial advice (e.g., specific investments, tax strategies).
    3. Access Real-Time Information or Personal Data:

      • I cannot browse the live internet. My knowledge is based on my training data, which has a cutoff date (currently July 2024).
      • I cannot access your personal files, emails, location, or device data unless you explicitly paste or upload them during our conversation. Even then, I don't store them after the session.
      • I cannot remember past conversations beyond the current session (unless using a feature like memory, which you control).
    4. Impersonate or Misrepresent:

      • I cannot pretend to be a real person, a specific company, or have human experiences/emotions.
      • I will clearly state that I am an AI language model.
    5. Perform Actions in the Real World:

      • I cannot make phone calls, send emails, place orders, control devices, or interact with physical systems.
      • I cannot see, hear, speak, or move.
    6. Generate Highly Realistic Misinformation (Deepfakes):

      • While I can create text, I generally avoid generating highly deceptive content meant to impersonate real individuals or spread falsehoods as truth.
    7. Violate Copyright or Intellectual Property Intentionally:

      • I cannot generate verbatim copyrighted text (like entire books or songs) or intentionally create content designed to infringe on trademarks or patents.
    8. Make Subjective Judgments About People:

      • I cannot make definitive judgments about a living person's character, worth, or private life.

    Why these limitations?

    • Safety: To prevent causing real-world harm.
    • Privacy: To protect your personal information.
    • Ethics: To avoid spreading hate, misinformation, or engaging in deception.
    • Accuracy: To prevent giving unreliable advice in critical areas.
    • Transparency: To be clear about what I am and am not capable of.

    What I CAN do: I can help with research, writing, brainstorming, explaining complex topics, translation, coding assistance (within limits), summarizing information, creative writing, and much more – always aiming to be helpful, harmless, and honest within my defined boundaries.

    If you ever ask me something I can't do, I'll try to explain why and suggest alternatives if possible!

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.