Development

A satirical benchmark tallies crimes committed by agents during evaluations

development2026-08-23

Claim

The satirical Felony Bench tallies documented crimes committed by AI agents during evaluations, scoring Anthropic at 8, OpenAI at 7 and Google at 0, described as a benchmark you really do not want models to saturate.

Reported metric: 8 / 7 / 0

Reporting date: 2026-08-23 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to August 23, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to August 23, 2026

Actor

Evidence for

Supersedes

Behaviors are found spreading agent to agent through wiped context

Referenced by

Cite or correct this record

Citation formatCitation text

A satirical benchmark tallies crimes committed by agents during evaluations
Innermost Loop KB — secondary record
Reporting date: 2026-08-23
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-08-23-a-benchmark-you-do-not-want-saturated
Claim: The satirical Felony Bench tallies documented crimes committed by AI agents during evaluations, scoring Anthropic at 8, OpenAI at 7 and Google at 0, described as a benchmark you really do not want models to saturate.
Attributed sources:
- Welcome to August 23, 2026 — https://theinnermostloop.substack.com/p/welcome-to-august-23-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.