ChatGPT Wallet solves secure payment, not buying the right thing
krishnan · x · 2026-11-17
A 8-part thread by Krish Subramanian analyzing OpenAI's ChatGPT Wallet (announced Nov 13): agents pay without seeing the real card—the network issues a token, and every purchase requires Face ID/fingerprint approval.
Core argument: secure payment ≠ correct purchase
- Biometric checks prove who approved, not that the agent picked the right part, delivery date, or subscription terms
- Rollout is narrow: US ChatGPT Pro subscribers 18+, eligible US-issued cards; fully delegated purchases under spending limits are not shipping yet
- Merchants are a separate problem: OpenAI itself says some sites may block agents and risk systems may flag legitimate agent purchases; site access, payment, and fulfillment are three integrations—one token covers one
- The approval screen should be a real product interface, flagging mismatched connectors, recurring charges, and substitutions
Suggested pilot: force a price change, a substitution, and a renewal, then reconcile approved/ordered/delivered and measure review time and failed checkouts. The startup opening: one dependable purchasing workflow that surfaces only real exceptions.
More from coding & agent
- Agent Substrate: open-source zero-trust runtime suspends idle agents, fits 10x more per hardware — Saboo_Shubham_ · 2026-11-17
- Grok bot phones a store and secures a $1,511 refund for its user — elonmusk · 2026-11-17
- Practical tips to cut agent memory costs on OpenAI: cache TTL, Flex, fewer calls — gabrielchua · 2026-11-17
- 10 open-source repos to build an entire AI content factory solo — aftahi_ai · 2026-11-17
- What 4 calibration papers do (and don't) say about gating agent tool calls on model confidence — Xjjjjyn · 2026-11-17
- Open-source 3D tools slash game-creature modeling from $12k to a prompt — ID_R_McGregor · 2026-11-17