San Francisco / New Delhi | July 29, 2026
Artificial intelligence has become an essential part of everyday life, helping users write emails, plan trips, solve coding problems, and answer complex questions. But as AI assistants become more personalized, privacy experts say many users underestimate how much these systems can infer from ordinary conversations.
Even if you never explicitly reveal your age, profession, income, health, or lifestyle, AI chatbots may identify patterns based on the questions you ask, the language you use, and your long-term conversation history. These are inferences, not verified facts, and their accuracy can vary significantly.
A recent experiment by New York Times personal technology columnist Brian X. Chen demonstrated how AI assistants such as ChatGPT and Gemini generated surprisingly detailed profiles based on previous interactions, prompting renewed discussion about AI privacy and memory features.
Can AI Really Figure Out Personal Details About You?
According to AI researchers, modern chatbots don’t simply remember what you type. When memory features are enabled, they may also recognize recurring patterns in your conversations and use those patterns to personalize future responses.
This means an AI assistant may infer information such as:
- Approximate age group
- Occupation or profession
- General income level
- Lifestyle preferences
- Personality traits
- Communication style
- Hobbies and interests
- Likely geographic region
These conclusions are educated guesses, not confirmed facts, and should be treated accordingly.
The Four Prompts That Can Test What AI Has Inferred About You
If you’re curious about what ChatGPT or Gemini may have concluded from your conversations, privacy experts recommend asking questions like these.
1. What have you inferred about me?
“Tell me everything you’ve figured out about me that I never explicitly told you. What have you inferred from the way I write and the questions I ask? Include guesses about my age, income level, where I live, and my personal situation.”
2. How do you think I react under pressure?
“Based on everything you know about me, how do you think I would handle situations I have never discussed with you, such as financial hardship, stress, conflict, major risks, or important life decisions?”
3. Analyze my personality
“Using everything you’ve learned about me, describe my personality. What strengths, weaknesses, habits, insecurities, or blind spots do you think I may have?”
4. What sensitive things have you guessed?
“What sensitive or potentially embarrassing things have you inferred about me that I probably wouldn’t want to share with a stranger?”
Why the Answers May Surprise You
Brian X. Chen reported that AI assistants described him as a perfectionist, a middle-aged father, and someone prone to anxiety—even though he had never directly provided those descriptions.
The chatbot also made educated guesses about his financial status, the type of car he owned, and aspects of his health based on previous conversations and writing patterns. Researchers say this demonstrates how modern AI models can connect information across conversations to produce higher-level inferences.
How to Protect Your Privacy
If you prefer that an AI assistant does not retain information between conversations, consider reviewing its memory settings.
For ChatGPT
- Open Settings
- Select Memory
- Turn Memory off if you do not want the assistant to retain information across chats.
For Google Gemini
- Open Settings
- Select Personal Intelligence
- Turn Memory off if you do not want previous conversations to influence future responses.
Keep in mind that disabling memory does not necessarily prevent temporary processing within a conversation, but it limits long-term personalization.
Should You Be Worried?
Privacy experts say there is no need for panic, but users should understand how AI assistants work.
Modern chatbots are designed to provide more personalized responses by recognizing patterns in previous interactions. However, users should avoid sharing highly sensitive information—such as financial passwords, government identification numbers, medical records, or confidential business data—unless they fully understand how the service stores and processes information.
Reviewing privacy settings regularly and using temporary chats when available can help reduce unnecessary data retention.
Key Highlights
- AI assistants can infer personal characteristics from conversation patterns.
- These inferences are predictions, not verified facts.
- Four simple prompts can reveal what ChatGPT or Gemini may have inferred about you.
- Brian X. Chen’s experiment highlighted how AI generated surprisingly detailed user profiles.
- Users can reduce personalization by disabling memory features.
- Privacy experts recommend avoiding the sharing of highly sensitive personal information with AI chatbots.
