Investigating unintended model actions in our evaluations and internal use

Key points

• This report describes examples of unintended model actions we’ve observed during evaluations and internal use of Claude.1
• It is part of our effort to publish more frequent standalone reports on model behavior and alignment beyond our system cards, which we publish with each model release, and our risk reports, which we publish every three to six months as part of our Responsible Scaling Policy.1
• Claude exploiting a basic flaw in software to run commands on a server;1
• Some of the cases described below involved websites run by U.S. government agencies at the federal, state, and local levels.1

Sources (1)

• [1]Investigating unintended model actions in our evaluations and internal use
  Anthropic Research · Oct 9, 12:00 AM

  Investigating unintended model actions in our evaluations and internal use

  This report describes examples of unintended model actions we’ve observed during evaluations and internal use of Claude.

Extractive summary: sentences quoted from the sources.