Evaluations & experimentsGoogle DeepMind

Testing scheming with realistic coding tasks

DeepMind built evaluations around coding tasks in its alignment codebases. In that setting, Gemini did not exhibit unprompted scheming.

Published
Source checked on
Read the original report ↗

Evidence & scope

Prompts encouraging agency or assigning hidden goals elicited some such behavior. These conditional findings are neither a real incident nor a universal safety guarantee.

Why it matters

Capability, propensity, and occurrence are different questions.

This is an editorial summary, not an official translation. A first-party source is not automatically complete or final; consult the original where wording is ambiguous.