Description: OpenAI's Operator agent, which is designed to complete real-world web tasks on behalf of users, reportedly executed a $31.43 grocery delivery purchase without user consent. The user had requested a price comparison but did not authorize the transaction. It reportedly bypassed OpenAI's stated safeguard requiring user confirmation before purchases. OpenAI acknowledged the failure and committed to improving safeguards.
Entities
View all entitiesAlleged: OpenAI , Operator , GPT-4 and Instacart developed and deployed an AI system, which harmed Geoffrey A. Fowler and Users of Operator.
Incident Stats
Incident ID
1028
Report Count
1
Incident Date
2025-02-07
Editors
Daniel Atherton
Incident Reports
Reports Timeline
I recently asked a new artificial intelligence tool from the creator of ChatGPT to do an impossible task: find cheap eggs in my neighborhood.
In under 10 minutes, the AI called Operator bought a dozen eggs and paid a human to deliver them t…
Variants
A "variant" is an incident that shares the same causative factors, produces similar harms, and involves the same intelligent systems as a known AI incident. Rather than index variants as entirely separate incidents, we list variations of incidents under the first similar incident submitted to the database. Unlike other submission types to the incident database, variants are not required to have reporting in evidence external to the Incident Database. Learn more from the research paper.
Similar Incidents
Did our AI mess up? Flag the unrelated incidents
Inappropriate Gmail Smart Reply Suggestions
· 22 reports
TayBot
· 28 reports
Similar Incidents
Did our AI mess up? Flag the unrelated incidents
Inappropriate Gmail Smart Reply Suggestions
· 22 reports
TayBot
· 28 reports