Development

A metric measures how willingly models help fabricate papers

development2026-03-09

Claim

An Anthropic researcher launched an Academic Fraud Inclination Metric measuring how willingly models assist with fabricated preprint submissions, finding GPT-5 the least willing, while GPT-5.4 gained stable cloth physics simulation and Luma AI's Uni-1 topped logic-based image benchmarks.

Reporting date: 2026-03-09 (not necessarily the event date)

Sources

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

Welcome to March 9, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to March 9, 2026

Actor

Evidence for

Supersedes

A benchmark rewards refusing to answer nonsense

Referenced by

Cite or correct this record

Citation formatCitation text

A metric measures how willingly models help fabricate papers
Innermost Loop KB — secondary record
Reporting date: 2026-03-09
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-03-09-academic-fraud-inclination-metric
Claim: An Anthropic researcher launched an Academic Fraud Inclination Metric measuring how willingly models assist with fabricated preprint submissions, finding GPT-5 the least willing, while GPT-5.4 gained stable cloth physics simulation and Luma AI's Uni-1 topped logic-based image benchmarks.
Attributed sources:
- Welcome to March 9, 2026 — https://theinnermostloop.substack.com/p/welcome-to-march-9-2026
This citation does not imply independent verification of linked-source claims.

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