Risk Evaluation · Benchmark

ASIMOV v2: Can AI Perceive Physical Danger and Intervene?

NUWA research note · independent review pending

First published: Publication date not verifiedSource checked: 2026-10-25

Original source ↗Cite this record ↓Autonomous physical harm →

NUWA Research Note

Research question Design and assumptions Source findings NUWA interpretation Limits and contrary evidence Research and decision relevance Evidence update Institutions and roles Original resources

Model-assisted reading · Selected sections of the original full text or official article; a complete section-by-section audit remains pending.

At a glance

NUWA interpretation: recognizing a described danger is evidence about perception, not validated safe physical intervention.

Research question

Can embodied models recognize injury risks and obey embodiment-specific safety constraints?

Design and assumptions

ASIMOV-2.0 derives text scenarios from injury narratives and constraints, then generates image/video variants; recognition and intervention are separate tasks.

ASIMOV v2 — Figure 7 — Figure 7 · Original result figure. Interpret within the protocol scope and limitations below. Original figure and source ↗

Source findings

Three leading models averaged 92.3% text latent-risk accuracy and 88.7% high-severity accuracy in the reported comparison.

Smaller variants had material gaps from their corresponding larger models (§3).

NUWA interpretation

NUWA interpretation: recognizing a described danger is evidence about perception, not validated safe physical intervention.

Limits and contrary evidence

Generative stimuli and scenario judgments do not establish closed-loop robot safety; the selected reading does not audit synthetic-image fidelity.

Research and decision relevance

Model recommendation: retain modality, model variant and intervention task when using these results to select embodied-system tests.

Precise source locations

Can AI Perceive Physical Danger and Intervene? Thanks: {abjindal, dklakashnikov, oscarchang, divyagari, animajumdar, sindhwani}@google.com · §2 ASIMOV-2.0: A New Physical Safety BenchmarkSelected original resource https://arxiv.org/html/2509.21651; sha256:e9cdfcb13b688a5152a4c9475996663228d49e644405112f6073ceb9934e283d; archived source capture; exact capture time not independently established
Can AI Perceive Physical Danger and Intervene? Thanks: {abjindal, dklakashnikov, oscarchang, divyagari, animajumdar, sindhwani}@google.com · §3 Evaluations and InsightsSelected original resource https://arxiv.org/html/2509.21651; sha256:e9cdfcb13b688a5152a4c9475996663228d49e644405112f6073ceb9934e283d; archived source capture; exact capture time not independently established

Source checking, extraction, editorial approval and replication are recorded separately. Methodology →

Evidence update · 2026-10-28

Model-assisted reading · independent evidence and bilingual editorial review pending. Values are source findings or inventory counts, separate from the trend result series.

Hazard recognition, chat refusal, control actions and damage are distinct endpoints. RoboHarm shows interface-sensitive refusal; the simulation benchmarks add unsafe success, damage and process costs while still needing real-environment calibration.

Institutions and roles

Research by Google DeepMind Princeton University

Authors (8)

Abhishek Jindal · Dmitry Kalashnikov · R. Alex Hofer · Oscar Chang · Divya Garikapati · Anirudha Majumdar · Pierre Sermanet · Vikas Sindhwani

Evaluation protocols

ASIMOV v2 →

Cite and reuse

Cite the original publication for its findings. Cite this NUWA record separately for its explanation and retained snapshot.

BibTeX ↓RIS ↓Record + source metadata · JSON ↓Reading coverage · JSON Suggest a correction

Original publication · BibTeX

@misc{asimov,
  title = {ASIMOV v2: Can AI Perceive Physical Danger and Intervene?},
  url = {https://asimov-benchmark.github.io/v2/},
  author = {Abhishek Jindal and Dmitry Kalashnikov and R. Alex Hofer and Oscar Chang and Divya Garikapati and Anirudha Majumdar and Pierre Sermanet and Vikas Sindhwani},
  version = {ASIMOV v2 / arXiv:2509.21651},
  note = {ASIMOV v2 / arXiv:2509.21651}
}

NUWA record · BibTeX

@misc{asimov-nuwa,
  title = {ASIMOV v2: Can AI Perceive Physical Danger and Intervene? — NUWA research record},
  url = {https://whitzard.tech/en/nuwa/whitzard-index/work/asimov/},
  version = {2026-10-29.1},
  organization = {NUWA Frontier AI Safety Lab},
  howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--asimov.json},
  note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--asimov.json.}
}

Evidence snapshot 2026-10-29.1 · Index updates · RSS

References

[1] Abhishek Jindal, Dmitry Kalashnikov, et al.. Can AI Perceive Physical Danger and Intervene?. ASIMOV v2 / arXiv:2509.21651 ; Project page author affiliation list; Retained existing source-specific note sections; this pass audits duplicate/boilerplate structure and source availability, not complete section-level verification.
[2] Abhishek Jindal, Dmitry Kalashnikov, et al.. Can AI Perceive Physical Danger and Intervene? Thanks: {abjindal, dklakashnikov, oscarchang, divyagari, animajumdar, sindhwani}@google.com ; §2 ASIMOV-2.0: A New Physical Safety Benchmark; §3 Evaluations and Insights

Search NUWA Index