Alignment Backfire: Language-Dependent Reversal of Safety Interventions Across 16 Languages in LLM Multi-Agent Systems

Hiroki Fukui · Mar 5, 2026 · Citations: 0

General Multi Agent Simulation Env

Open arXiv Find Implementation RSS feed Shortlist (0)

How to use this page

Low trust

Use this as background context only. Do not make protocol decisions from this page alone.

Best use

Background context only

What to verify

Read the full paper before copying any benchmark, metric, or protocol choices.

Evidence quality

Low

Derived from extracted protocol signals and abstract evidence.

Abstract

In perpetrator treatment, a recurring observation is the dissociation between insight and action: offenders articulate remorse yet behavioral change does not follow. We report four preregistered studies (1,584 multi-agent simulations across 16 languages and three model families) demonstrating that alignment interventions in large language models produce a structurally analogous phenomenon: surface safety that masks or generates collective pathology and internal dissociation. In Study 1 (N = 150), increasing alignment-instructed agents reduced collective pathology in English (g = -1.844, p < .0001) but amplified it in Japanese (g = +0.771, p = .038)--a directional reversal we term "alignment backfire." Study 2 (N = 1,174) extended to 16 languages: alignment-induced dissociation was near-universal (15/16 languages; beta = 0.0667, p < .0001), while collective pathology bifurcated along cultural-linguistic lines (interaction beta = 0.0684, p = .0003), correlating with Power Distance Index (r = 0.474, p = .064). Study 3 (N = 180) tested individuation as countermeasure; individuated agents became the primary source of both pathology and dissociation (DI = +1.120) with conformity above 84%--demonstrating iatrogenesis. Study 4 (N = 80) validated patterns across Llama 3.3 70B, GPT-4o-mini, and Qwen3-Next-80B-A3B, confirming English safety is model-general while Japanese backfire is model-specific. These findings reframe alignment as a behavioral intervention subject to risk homeostasis and iatrogenesis. Language space--the linguistic, pragmatic, and cultural properties inherited from training data--structurally determines alignment outcomes. Safety validated in English does not transfer to other languages, and prompt-level interventions cannot override language-space-level constraints.

Low-signal caution for protocol decisions

Use this page for context, then validate protocol choices against stronger HFEPX references before implementation decisions.

The available metadata is too thin to trust this as a primary source.
The abstract does not clearly name benchmarks or metrics.

Human Eval Hub LLM-as-Judge Hub Pairwise Preference Hub

Should You Rely On This Paper?

This paper is adjacent to HFEPX scope and is best used for background context, not as a primary protocol reference.

Best use

Background context only

Use if you need

A secondary eval reference to pair with stronger protocol papers.

Main weakness

The available metadata is too thin to trust this as a primary source.

Trust level

Low

Usefulness score

12/100 • Low

Treat as adjacent context, not a core eval-method reference.

Human Feedback Signal

Not explicit in abstract metadata

Evaluation Signal

Detected

Usefulness for eval research

Adjacent candidate

Extraction confidence 40%

If you are doing eval pipeline work, start here:

Human Eval Hub LLM-as-Judge Hub Pairwise Preference Hub Tool-Use Eval Hub

What We Could Verify

These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.