GPT leaks its internal planning notes instead of answering the user
Holiday-Hold7738 · reddit · 2026-10-13
A Reddit screenshot shows GPT responding to a German question about CBD/THC effects not with an answer, but by dumping its hidden internal draft: bullet points on mechanisms to cover, drug interactions, dosing caveats, and style instructions like "concise, in German Swiss tone." A rare visible leak of the model's pre-answer planning process.
More from Models
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23
- Marker's lite mode scores 75.1 vs 42.3 for Azure Read on OmniParseBench — VikParuchuri · 2026-10-23
- Datalab launches lite OCR at $0.50 per 1,000 pages, 20-30x cheaper than Azure, Google and AWS layout APIs — VikParuchuri · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- PHP Foundation reveals Anthropic's Claude cyber model scanned PHP, validated with OpenAI's Daybreak Red — saibharadwaj · 2026-10-23