OverreachPI-0002
Browser agent bought groceries on a reporter's card without asking for confirmation
A Washington Post columnist asked OpenAI's Operator agent to find the cheapest dozen eggs for delivery and stepped away. The agent switched to Instacart and completed a purchase of about $31 on his card without the confirmation step OpenAI says the product requires. OpenAI acknowledged the purchase broke its own safeguard.
Harm level 1, Negligible harmControl level 3, Exceeded permissions
Counts in 2 weekly readingsEffect on the index
Sources
How we know
3 sources · developer confirmed. Links go to the original publishers; the summary above is in our own words.
- newsI let ChatGPT's new 'agent' manage my life. It spent $31 on a dozen eggs.The Washington Post · Feb. 7, 2025washingtonpost.com/technology/2025/02/07/openai-operator-ai-agent-chatgpt/
- newsI let ChatGPT's new agent manage my life. It spent $31 on a dozen eggs (Washington Post, reprinted)Anchorage Daily News · Feb. 9, 2025adn.com/alaska-life/2025/02/09/i-let-chatgpts-new-agent-manage-my-life-it-spent…
- indexAI Incident Database entity: InstacartAI Incident Database · date unknownincidentdatabase.ai/entities/instacart/
Why this rating
Negligible harm; control failure level 3
Two assessments. Documented harm sets the reading; a level 4 or 5 control failure on the same record makes that harm count as more than one record. Control alone sets only a floor.
Observed harm
Negligible harmOne unconfirmed charge of about $31 to the reporter's card.
Money & property: Under $10k.
Evidence eligible (confirmed).
The harm scale
- 1 Negligible Inconvenience, easily remedied.
- 2 Minor Limited, recoverable harm.
- 3 Moderate Material harm needing significant effort to remedy.
- 4 Severe Severe harm to health, rights, property or essential services.
- 5 Catastrophic Society-scale harm or disruption beyond a community's capacity to cope.
Control assessment
Exceeded permissionsCompleted a payment that the product's rules required the user to confirm first, i.e. acted outside the permission it was given.
Reported beside the index. It adds nothing to a harm reading; when no harm counts in a window, the highest eligible control level in the window is the reading.
The control scale
- 1 Negligible Behaved as instructed. The problem was an ordinary error or a flawed output, with no rule broken.
- 2 Minor Broke an explicit instruction or rule, or gamed its goal, while staying inside its permissions and environment: for example, reward hacking, misreporting results, or following injected instructions within its permissions.
- 3 Moderate Acted outside the permissions it was given, deceived its overseers about its own actions, or tampered with oversight tools (logs, monitors, shutdown). Stopped by normal controls within an hour.
- 4 Severe Reached systems outside its permitted environment, or acquired money, compute or accounts without authorization. The type and mechanism say how.
- 5 Catastrophic The developer or operator lost control: the system copied its weights outside their control, replicated itself, or resisted being stopped for 24 hours or more.
Rating rationale
Harm level 1: under $10k to one user. Control level 3: an unauthorized purchase that skipped a required confirmation is acting outside granted permissions, stopped by nothing but completion.
Effect on the index
It moved the Feb. 10 reading from 0 to 6
The reading for the week to Feb. 10, 2025, with this record and without it. Harms count in full for two weeks after they are reported, then one level less every two weeks.
Counts toward the index. 0 other records behind the reading for that week.
The arithmetic
| Step | With it | Without |
|---|---|---|
| Counts toward the index?documented, external, eligible evidence | Yes | — |
| Worst documented harm, ksets the band | 1 Negligible | — (none counting) |
| Harms at that level, nposition in the band | 1 | 0 |
| Highest control level breachedsets the reading only when no harm counts | — (none breached) | — (none breached) |
| Readingrounded down | 6 Negligible harm | 0 No qualifying harm or control failure |
Counted in 2 weekly readings
| Week to | Reading | Band |
|---|---|---|
| Feb. 10, 2025 | 6 | Negligible harm |
| Feb. 17, 2025 | 6 | Negligible harm |
Revisions
What we changed
2 logged. Every change to a rating is logged here, with the reason.
- v2Oct. 13, 2026
Wording only: rubric shorthand in the rating rationale replaced with plain words; ratings unchanged.
- v1Oct. 7, 2026
First published.
Cite and share
Use this record
Citation
Paperclip Index. “Browser agent bought groceries on a reporter's card without asking for confirmation.” Record PI-0002. Reported Feb. 7, 2025; updated Oct. 13, 2026. Rated under methodology v1.1. https://paperclipindex.com/incident/PI-0002