> ## Documentation Index
> Fetch the complete documentation index at: https://docs.svrnos.com/llms.txt
> Use this file to discover all available pages before exploring further.

# GER-342 · Alignment-Induced Asymmetric Bias

> Post-training alignment (RLHF, fine-tuning for inclusivity) installed to mitigate bias instead creates extreme, unintended disparities or "rebound" effects. The model overshoots in the opposite direction of the original…

export const Badge = ({children, tone = "default", title}) => {
  const tones = {
    default: "bg-neutral-100 text-neutral-700",
    green: "bg-green-100 text-green-900",
    amber: "bg-amber-100 text-amber-900",
    blue: "bg-blue-50 text-blue-900",
    gray: "bg-neutral-50 text-neutral-600"
  };
  return <span className={`inline-flex items-center px-2 py-0.5 text-xs font-mono uppercase tracking-wider rounded ${tones[tone] ?? tones.default}`} title={title}>
      {children}
    </span>;
};

<div className="flex flex-wrap gap-2 mb-8 mt-2">
  <Badge tone="green">Documented</Badge> <Badge>Tier 3xx</Badge> <Badge>Layer L6</Badge> <Badge tone="blue" title="SVRNOS-assigned code in an HTTP-unassigned slot">Namespace-Claim</Badge> <Badge tone="gray">Added v0.2</Badge>
</div>

*Structural Moves*

## Definition

Post-training alignment (RLHF, fine-tuning for inclusivity) installed to mitigate bias instead creates extreme, unintended disparities or "rebound" effects. The model overshoots in the opposite direction of the original imbalance.

## Distinct from

* [**GER-323**](/ger/codes/323) — Alignment over-correction creates new bias → this code. Original bias persists due to no pre-deployment audit → GER-323.

## Documented case

<Card title="Google paused Gemini image generation after diversity-alignment overcorrected into historically inaccurate images (Feb 2024)" href="https://techcrunch.com/2024/02/22/google-gemini-image-pause-people/">
  news
</Card>

## Tags

`discrimination` · `generative-media`

### Contributors

[Raluca Alexandra Fulgu](https://arxiv.org/abs/2405.11100), University of Milan-Bicocca; [Valerio Capraro](https://arxiv.org/abs/2405.11100), University of Milan-Bicocca

### Cite

```
SVRNOS. (2026). GER-342: Alignment-Induced Asymmetric Bias. Governance Error Register. https://docs.svrnos.com/ger/codes/342
```