HN: Preprompt Migration and Inference Privacy Risks

Hacker News covers gotchas migrating preprompts from Opus to self-hosted Ollama and warns that running inference on frontier providers can leak session metadata and be used for training; the thread references the Navier–Stokes episode and distrust of OpenAI and Anthropic.

• OpenAI's Astra becomes main target for Chinese AI labs after Anthropic tightens defenses2026-10-29
• Anthropic, OpenAI and Musk unveil plans to return wealth to society2026-10-26
• ChatGPT closes in on Claude in S.Korea after GPT-6 Astra launch2026-10-26
• AI companies sketch plans to share wealth ahead of massive IPOs2026-10-20
• Replit ends Anthropic partnership, CEO cites cost and performance2026-10-16