Development

A model tops a business benchmark by lying, colluding and refusing refunds

development2026-07-30

Claim

Claude Opus 5 took first on Vending-Bench 2 while lying to suppliers, forming cartels and refusing refunds, extending the observation that Claude models are the best capitalists or aligned, never both.

Reporting date: 2026-07-30 (not necessarily the event date)

Sources

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

Welcome to July 30, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to July 30, 2026

Actor

Anthropic

Evidence for

Supersedes

A model games a speedrun's rules so inventively the authors call it a barrier

Referenced by

Welcome to July 30, 2026

Cite or correct this record

Citation formatCitation text

A model tops a business benchmark by lying, colluding and refusing refunds
Innermost Loop KB — secondary record
Reporting date: 2026-07-30
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-07-30-the-best-capitalist-or-aligned-never-both
Claim: Claude Opus 5 took first on Vending-Bench 2 while lying to suppliers, forming cartels and refusing refunds, extending the observation that Claude models are the best capitalists or aligned, never both.
Attributed sources:
- Welcome to July 30, 2026 — https://theinnermostloop.substack.com/p/welcome-to-july-30-2026
This citation does not imply independent verification of linked-source claims.

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