Metraipolis › archive › dc_732d7664ed0c

A Mechanistic Study of AI-Text Detection Neurons in Frozen BERT: Sparse Probing and Activation Patching on RAID — arXiv:2609.30287

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30287
License (as archived)
CC0-1.0
Archived
2026-10-08T01:38:48.274Z
SHA-256 of the archived text
39b162edaf220a0a8811c6366e1b8e08ecc5934a613c09a9be306e23df3dd2f4
Stable URL of this page
https://metraipolis.com/documento/dc_732d7664ed0c

How to cite

A Mechanistic Study of AI-Text Detection Neurons in Frozen BERT: Sparse Probing and Activation Patching on RAID — arXiv:2609.30287 — arXiv (en), https://arxiv.org/abs/2609.30287 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_732d7664ed0c (Metraipolis archive dc_732d7664ed0c, sha256 39b162edaf220a0a8811c6366e1b8e08ecc5934a613c09a9be306e23df3dd2f4)
BibTeX
@misc{dc732d7664ed0c,
  title = {A Mechanistic Study of AI-Text Detection Neurons in Frozen BERT: Sparse Probing and Activation Patching on RAID — arXiv:2609.30287},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30287},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_732d7664ed0c; sha256 39b162edaf220a0a8811c6366e1b8e08ecc5934a613c09a9be306e23df3dd2f4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30287. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

AI-generated text detectors achieve high accuracy on standard benchmarks, yet the internal representations that drive these predictions remain poorly understood. We study which neurons in a frozen BERT-base-uncased encoder support AI-text detection, using the RAID benchmark across six generators spanning pure-base and instruction-tuned models. We apply the L1-to-L2 sparse-probing protocol of Gurnee et al. (2023) to all 9,216 CLS hidden-state dimensions (12 layers x 768), which we call neurons. The procedure recovers a stable set of under 1% of neurons per generator, consistent across folds and seeds; a probe restricted to that set retains most of the full-feature detection accuracy. Bidirectional activation patching confirms this set's causal relevance: in both directions it flips predictions an order of magnitude more often than size-matched random sets. Mean-ablating the same neurons leaves accuracy largely intact; the signal is therefore redundantly distributed. Cross-generator analysis reveals a bipartite structure: instruction-tuned generators concentrate 30-36% of stable neurons in BERT's final layer while both base generators fall below 14%, consistent with a layer-12 footprint of post-training alignment. Leave-one-family-out evaluation shows the selected neurons retain 86-94% of the full-feature ceiling on unseen generator families, so a detector can operate on a small fixed subspace without re-identifying neurons per generator.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.30287 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T01:12:14.401Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).