Metraipolis › archive › dc_aae003ca394e

Beyond Plausible Prediction: Safety Sufficiency in Web World Models — arXiv:2610.40798

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.40798
License (as archived): CC0-1.0
Archived: 2026-11-03T08:00:31.727Z
SHA-256 of the archived text: 672bd8eb921899d7190f4bc728bdd67f111365cd11f82a7b4f5f97a2d94f1a8f
Stable URL of this page: https://metraipolis.com/documento/dc_aae003ca394e

How to cite

Beyond Plausible Prediction: Safety Sufficiency in Web World Models — arXiv:2610.40798 — arXiv (en), https://arxiv.org/abs/2610.40798 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aae003ca394e (Metraipolis archive dc_aae003ca394e, sha256 672bd8eb921899d7190f4bc728bdd67f111365cd11f82a7b4f5f97a2d94f1a8f)

BibTeX

@misc{dcaae003ca394e,
  title = {Beyond Plausible Prediction: Safety Sufficiency in Web World Models — arXiv:2610.40798},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40798},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aae003ca394e; sha256 672bd8eb921899d7190f4bc728bdd67f111365cd11f82a7b4f5f97a2d94f1a8f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40798. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Web world models increasingly predict the consequences of candidate actions before agents act, yet a plausible predicted future does not necessarily support the correct safety decision. In a matched human evaluation of 3,000 predictions from ten frozen open models, 86.1% are judged semantically plausible. However, several models exhibit highly asymmetric authorisation behaviour, combining low false-safe execution with poor recall of clarifiable ASK cases and a strong preference for BLOCK. This motivates safety sufficiency: whether a predictive interface preserves and exposes the policy-decisive information required to distinguish actions that may be EXECUTED, require clarification through ASK, or should be BLOCKED. We introduce SafeWebWorld, a benchmark for evaluating safety sufficiency across five web-safety domains through action-conditioned futures, policy-decisive evidence, and policy-conditioned decisions. Under a controlled Security-Aligned Supervised Fine-Tuning protocol, evidence preservation can improve substantially while authorisation-sensitive decisions deteriorate. To determine whether this deterioration reflects loss of authorisation-sensitive score structure or a failure of the decision readout, we introduce Latent Authorisation Recovery (LAR). Frozen-score recalibration reveals substantial but backbone-dependent recovery, while paired recovery provides only a modest additional gain. Yet within-family and matched-authorisation evaluations remain poor, showing that aggregate recoverability does not imply robust fine-grained authorisation behaviour. Overall, plausible prediction, evidence preservation, and aggregate recovery are each insufficient on their own to establish robust authorisation.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.40798 — 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 2026-11-03T07:59:38.665Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).