12 prompts to force ChatGPT to reveal — and delete — the psychological profile it builds on you
nikola_mr64990 · x · 2026-11-19
A viral X thread walks users through auditing and purging the personal profile ChatGPT silently accumulates: dump the raw memory list, ask the model to reveal everything it has inferred (age range, income, location, goals) with confidence and clues, surface blind-spot personality traits, let it predict your reactions to money/stress/risk, then delete memories via Settings → Customization → Memory and disable "Improve the model for everyone" under Data Controls so chats aren't used for training. It also suggests asking the model to rank everything it knows by privacy sensitivity first.
More from Safety
- AI safety groups afraid to speak out over Anthropic funding ties, researcher claims — arvindsatya1 · 2026-11-19
- Open-source playbook catalogs 117 App Store rejection patterns, gains 542 stars in a day — mjmirza · 2026-11-19
- Parody "Initial Extinction Offering" Prospectus Hands Out Class H Shares Outside Anthropic — gleech · 2026-11-19
- Palo Alto CEO: critical vulns 7x last year's average, AI-assisted attacks rising — Insomniac897 · 2026-11-19
- Axios exclusive: Bots amplified conspiracy theories and panic during recent tech outages — GaryMarcus · 2026-11-19
- European Commission hit by cyberattack, payroll and medical reimbursement systems down for ~10 days — GaryMarcus · 2026-11-19