Shared AI agent knows everyone's secrets: how do you plan group events without leaking them?
IndicationTypical972 · reddit · 2026-11-12
A Redditor highlights an under-discussed agent interaction problem: four friends privately tell the same AI their constraints (broke until payday, lab until 7pm Thursday, hates loud venues), then the group chat asks it to pick a dinner spot. The AI holds information that would be awkward if leaked — either by stating 'Priya can't afford it' or by quietly never suggesting pricey options.
Per the project's page, it checks every group reply against each person's private disclosures and strips out giveaways, enforced in code rather than prompt instructions; friends only hear whether a time works. The privacy feature only activates in groups of three or more, since in a pair the constraint's source is obvious. The author doubts this holds in practice — inference is hard to stop — and asks whether shared multi-user agent context can ever avoid leaking through behavior.
More from coding & agent
- antirez: AI code generation now works when task boundaries are clear — antirez · 2026-11-12
- One-prompt 3D Solar System: Qoder's new Qwen4-Max-Preview builds an explorable browser universe — hey_abusiddik · 2026-11-12
- Giving agents a whiteboard: Claude Code crew plans on a shared Excalidraw canvas for $0.36 — pramodk73 · 2026-11-12
- AI agents automate the grunt work engineers learn from — how to train the next generation? — bendee983 · 2026-11-12
- KEEL, an Open Standard for AI Coding Project Docs, Launches keel.to — DominusFL · 2026-11-12
- PyTorch 2.15 regression slows LU solver up to 100x, warns researcher — ducha_aiki · 2026-11-12