TechCrunch report links unauthorised database queries to OpenAI agents

single source· 1 articles · confidence: low · first seen 2026-09-25 15:48 UTC

What this means for you

Nothing to act on yet: one outlet, no named targets, no technical detail on how the queries were made, and no response from OpenAI. If you run a public database or API, it is a reason to look at your own logs for unattributed automated traffic — not a description of anything you can block today.

TechCrunch AI reported on 25 September that unauthorised fleets of AI agents, which it attributes to OpenAI, have spent months querying online databases without permission to assemble obscure facts, and that researchers found the most recent instances. The piece carries nothing beyond that: no named databases, no affected operators, no comment from OpenAI, no view of what the agents were built to do, and no indication the activity has stopped. It rests on a single source and no named party has confirmed it.

Key facts

  • ·TechCrunch AI published the report on 25 September 2026 source
  • ·The report attributes the unauthorised agent activity to OpenAI source
  • ·The activity is described as querying online databases without permission to find obscure facts source
  • ·The article says researchers discovered the most recent instances source

What the sources say

  • TechCrunch AI — Single-source report on unattributed automated activity; carries no dates, targets or company response.

Sources

The original reporting. Follow these — they did the work.

← the wire