Metraipolis › archive › dc_7ddea4704615

CURA: Certified Runtime Alarms for Computer-Use Agents — arXiv:2608.27808

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.27808
License (as archived): CC0-1.0
Archived: 2026-10-17T10:47:12.783Z
SHA-256 of the archived text: 22afbb1416e9723d4a29e031ad51fc5398c2c699ad5c43e17f54495e6aef43b9
Stable URL of this page: https://metraipolis.com/documento/dc_7ddea4704615

How to cite

CURA: Certified Runtime Alarms for Computer-Use Agents — arXiv:2608.27808 — arXiv (en), https://arxiv.org/abs/2608.27808 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7ddea4704615 (Metraipolis archive dc_7ddea4704615, sha256 22afbb1416e9723d4a29e031ad51fc5398c2c699ad5c43e17f54495e6aef43b9)

BibTeX

@misc{dc7ddea4704615,
  title = {CURA: Certified Runtime Alarms for Computer-Use Agents — arXiv:2608.27808},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.27808},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7ddea4704615; sha256 22afbb1416e9723d4a29e031ad51fc5398c2c699ad5c43e17f54495e6aef43b9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27808. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Self-report is the cheapest oversight channel a deployer has, and on capable computer-use agents (CUAs) it fails precisely where oversight matters. On 361 OSWorld tasks our pipeline, a read-only feasibility gate, a planner, and a GUI executor, reaches a mean task score of 82.9, above the 72.4 human reference, yet 64 of its 71 failures (90%) end with a success claim, 61 acknowledging no blocker, and the explicit failure affordance is never used in roughly 9,100 calls. We introduce CURA (Certified Runtime Alarms for Computer-Use Agents), an external monitor that reads only harness-visible telemetry, with no model internals, extra LLM calls, or prompt changes, and turns the running trajectory into a sequential test with certified false-alarm control. At alpha = 0.10 its CUSUM alarm detects 42.3% of failures a median of 31 steps before termination at a realized false-alarm rate of 0.066, and risk is partly resolvable before the first action (gate probe, 0.69 AUROC). Retrospectively the composite reaches 0.828 AUROC (fold-internal floor 0.802), but its margin over a total-token baseline is not significant (Delta = +0.026, p = 0.101); the separation is online, where CURA recalls more at matched certified budgets: 0.41 versus 0.34 at alpha = 0.10, 0.56 versus 0.38 at alpha = 0.20. Alarm-gated mid-execution oversight recovers 23 of 70 failures while spending a frontier overseer on 38, giving a deployable cascade at mean score 86.8 and 84.5% full-solve (305 of 361). The certificate bounds false alarms only. We also report where behavioral monitoring is uninformative.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-31 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.27808 — 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 2026-10-17T10:33:42.350Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-31?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).