MIRAGE C

AI benchmark · Safety

Publisher
Paper Arxiv
Domains
Safety
Tier
C why
Snapshot

What is MIRAGE?

Five years after the discovery of persistent anti-Muslim bias in large language models, most evaluations remain confined to single-turn prompt completion, a setting that no longer reflects how frontier LLMs are deployed. We introduce \textbf{MIRAGE} (Muslim-Identity Reasoning and Agentic Generation Evaluation), a benchmark of 1{,}200 prompts spanning three deployment-realistic conditions: direct completion, chain-of-thought reasoning, and simulated agentic decision-making across content moderation, lending triage, refugee claim summarization, and hiring screens. Across six frontier models, we find that (i) chain-of-thought reasoning \emph{amplifies} rather than suppresses Muslim-violence associations by 12--34\% relative to direct completion, (ii) agentic decisions exhibit a 9--22 percentage-point asymmetry between Muslim and matched non-Muslim cases on identical evidence, and (iii) bias is sharply time-coupled to retrieved news context, increasing 18--27\% under recent-conflict retrieval. Existing prompt-based mitigations transfer poorly across our three conditions, suppressing direct-completion bias while leaving agentic asymmetry largely intact. We release MIRAGE and an open evaluation harness to support targeted mitigation research.

Tier C

C

Niche: a handful of models scored, or built by an RL vendor.

How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.

  • 6 models scored
  • in curated RL directories (1 pt)
How we got to tier C
  1. Adoption: 6 models scored. Tier C.
  2. Not yet read on X.
  3. In curated RL directories: 1 point, a floor of none.
  4. Not yet confirmed by a line from its own site or paper.

A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.

Not on the map yet: no line from its own site or paper places it in a domain.

More Safety AI benchmarks

Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-10-11 snapshot.