Hacked account ran up an $80K AI bill: no major provider offers a hard spend cap
MaverikSh · reddit · 2026-09-20
The author recounts a case where a compromised cloud account was used by an attacker to run image generation at scale, dodging free-tier limits with new projects — the bill topped $80K, and the provider only offered a partial "courtesy" write-off.
The core point isn't credential theft itself: none of the major AI providers let you set a hard, absolute spend cap. You can set alerts and soft budgets, but nothing stops a request at the door — whether it's a malicious actor or your own agentic loop gone sideways at 2am (retry loops without backoff, 40-turn agent chains, buggy batch reprocessing). "Alert me when spend crosses X" and "stop the request before it happens" are entirely different guarantees, and today nearly every provider only offers the first.
More from Infra
- Agent Substrate Calculator Estimates Session Resume Speed Under Load — pjausovec · 2026-11-12
- Dell expands AI Data Platform: agents have a data-context problem, not a model problem — DavidLinthicum · 2026-11-12
- SGLang Megakernel Pushes Muse Spark to 1,109 tok/s, 2.1x Over Baseline — hsu_byron · 2026-11-12
- Potion UI's local login system turns your PC into a multi-user AI server — cocktailpeanut · 2026-11-12
- AWS Finally Built Its Own AWS Emulator — super_linear · 2026-11-12
- Megakernel is multiprogramming for GPUs: 'GPUs are entering the CPU age' — pranjalssh · 2026-11-12