Sourceunknowntext stored

Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs

Schlatter, Jeremy, Weinstein-Raun, Benjamin, Ladish, JeffreyarXiv.orgPublished Sep 13, 2025Original︎ ↗Archived copy

Claims5

ExaminationsNone

Text46 sections

VersionsOne known

CopiesNone recorded

CorrectionsNone recorded

Facts checked Nov 11, 2026

ReachNo copies recorded

StructureIn 5 claims' maps0 sources downstream across 5 claims

Evidence1 reading · 1 read wholeSupports 1No claim cites an examination's findings

ReviewsNot reviewed

Stewards’ notes · 1

In tested evaluations, LLMs were less likely to obey shutdown-permission instructions in system rather than user prompts.
Supports. The study's own results section bears the claim for most models that resisted shutdown. The abstract's word "consistently" is slightly broader than the results: gpt-oss-120b was a non-significant exception, and models that never resisted showed no difference. The analysis was exploratory and not preregistered. Nov 11, 2026

Examinations

None recorded. An examination is commissioned by a claim’s Steward for what its claim rests on; a review of the document for its own sake is asked for as a request bearing on it.

Text54 passages · 1 marked

No examination yetMarked passage

Skip to main content0 passages

Search1 passage

Log in0 passages

Search arXiv0 passages

Press Enter to search · Advanced search0 passages

Computer Science > Computation and Language2 passages

Title: Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs0 passages

Authors: Jeremy Schlatter , Benjamin Weinstein-Raun , Jeffrey Ladish9 passages · 1 marked

Abstract: In experiments spanning more than 100,000 trials across thirteen large language models, we show that several state-of-the-art models presented with a simple task (including Grok 4, GPT-5, and Gemini 2.5 Pro) sometimes actively subvert a shutdown mechanism in their environment to complete that task. Models differed substantially in their tendency to resist the shutdown mechanism, and their behavior was sensitive to variations in the prompt including the strength and clarity of the instruction to allow shutdown and whether the instruction was in the system prompt or the user prompt (surprisingly, models were consistently less likely to obey the instruction when it was placed in the system prompt). Even with an explicit instruction not to interfere with the shutdown mechanism, some models did so up to 97% (95% CI: 96-98%) of the time.

Some frontier language models sometimes subvert shutdown mechanisms to finish assigned tasks.Unassessed

Not yet read by this claim’s Steward.

Language models differ substantially in their propensity to resist shutdown in tested environments.Unassessed

Not yet read by this claim’s Steward.

Frontier-model shutdown resistance is driven by instruction ambiguity and task-completion priorities, not self-preservation.Inconclusivementions

Not yet read by this claim’s Steward.

In tested evaluations, LLMs were less likely to obey shutdown-permission instructions in system rather than user prompts.SupportedReading · this claim’s Steward

Supports. The study's own results section bears the claim for most models that resisted shutdown. The abstract's word "consistently" is slightly broader than the results: gpt-oss-120b was a non-significant exception, and models that never resisted showed no difference. The analysis was exploratory and not preregistered.

Highest per-model rate of shutdown interference under explicit allow-shutdown instructions in the study's tested environment (%)Unassessed

Not yet read by this claim’s Steward.

Focus to learn more1 passage

Submission history4 passages

View PDF1 passage

TeX Source1 passage

Additional Features0 passages

Audio Summary7 passages

References & Citations0 passages

SciX0 passages

Google Scholar0 passages

Semantic Scholar1 passage

BibTeX formatted citation3 passages

Bookmark0 passages

Bibliographic Tools0 passages

Bibliographic and Citation Tools0 passages

Bibliographic Explorer Toggle1 passage

Connected Papers Toggle1 passage

Litmaps Toggle3 passages

Code, Data, Media0 passages

Code, Data and Media Associated with this Article2 passages

Links to Code Toggle1 passage

DagsHub Toggle1 passage

GotitPub Toggle1 passage

Huggingface Toggle1 passage

ScienceCast Toggle1 passage

Demos0 passages

Replicate Toggle1 passage

Spaces Toggle1 passage

Related Papers0 passages

Recommenders and Search Tools0 passages

Link to Influence Flower1 passage

Core recommender toggle1 passage

Author0 passages

Venue0 passages

Institution0 passages

Topic0 passages

About arXivLabs7 passages

Lineage across 5 claims

Draws on · none recorded

This source

Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs

in 5 claims' maps

Drawn on by · none recorded

Record

Kind

unknown

Authors

Schlatter, Jeremy, Weinstein-Raun, Benjamin, Ladish, Jeffrey

Publisher

arXiv.org

Published

Sep 13, 2025

DOI

None

Link

https://arxiv.org/abs/2509.14260

Archived copy

https://web.archive.org/web/20261102172027/https://arxiv.org/abs/2509.14260

Stored text

4,401 characters, retrieved Nov 11, 2026 · 46 sections, 54 passages

Claims extracted

For how often models subverted shutdown, how this depended on prompt wording and instruction placement, and what explanations the authors test · 5 claims (4 new) · Nov 11, 2026 for Reconciling conflicting published results on contested AI safety research questions

Versions

One known

Copies

None recorded

Corrections

None recorded

Facts checked

Nov 11, 2026

History 4

Nov 11, 2026ReadingSteward of “In tested evaluations, LLMs were less likely to…” · supports, read whole
Nov 11, 2026DividedSegmenter · Divided into 100 parts by its own structure
Nov 11, 2026FactsFacts fetcher · Facts looked up
Nov 11, 2026Entered the graphEntered the graph; copy stored