Reddit survey hunts real stories of runaway AI agents: infinite retries and burned credits
masterai01 · reddit · 2026-09-25
A researcher is collecting real-world stories of runaway AI agents: repeated fixes, agents bouncing tasks between each other, endless retries, background sub-agents that never stop, and auto top-up silently refilling credits overnight.
The thread asks five questions: what kind of agent (coding, support, voice, workflow), self-built or vendor product (Claude Code, Cursor, LangGraph, Replit, etc.), measurable cost in dollars/credits/time and whether the vendor refunded, mitigations applied (spend caps, iteration limits, disabling auto top-up, monitoring), and what would fix it (better detection, working hard stops, usage visibility, refunds). The author says nothing is being sold and will post a summary of findings back to the community.
More from coding & agent
- Stripe engineer answers 'what's it like at Stripe' with demos over memos culture — schwentker · 2026-09-25
- Perplexity's Fast Search costs $1 per 1,000 requests at 200 ms, 5x cheaper than standard search — AravSrinivas · 2026-09-25
- Wasmer runs a real PostgreSQL 18.4 server on iOS and in the browser via WebAssembly — jedisct1 · 2026-09-25
- Google Cloud details 4 patterns for real-time AI voice agents that see, talk, think and code — rseroter · 2026-09-25
- Opus 5.5 coded a polished 30-second hype video from one prompt in 3 hours of autonomous work — DaveRogenmoser · 2026-09-25
- NeurIPS Oral: IDS co-evolves code with formal proofs, hitting 3x Claude Code's success rate — adityagp · 2026-09-25