Skip to content

Repository files navigation

STRESS-Test

tests paper license

Semantic-preserving Transformations for Robust Evaluation of Synthetic-text Screening

STRESS-Test is a research framework for evaluating whether AI-text detectors remain reliable after paraphrasing, translation, grammar correction, human editing, and character-level perturbations. Transformations are evaluated on both machine and human text at a fixed decision threshold, making evasion and false-accusation risk part of the same protocol.

Research status: active, pre-submission work. The current results are preliminary sensitivity analyses, not a finalized benchmark. Semantic-preservation thresholds still require the planned human-validation study, and the present per-domain calibration analysis estimates thresholds from the same clean domain slices used for evaluation. These limitations are stated here because they affect how the numbers should be interpreted.

Paper draft · Generated results · Reproducibility guide · Result provenance

Research questions

  1. Do clean-text accuracy and robustness rank detectors differently?
  2. How do realistic transformation pipelines affect detection at a frozen operating threshold?
  3. Do benign edits to human writing increase false-accusation rates for particular registers or writer groups?
  4. Which conclusions change when results are stratified by text difficulty, domain, and detector mechanism?

Preliminary results

The table below is generated from the current per-domain calibration sensitivity analysis at a target 1% false-positive rate. Robustness Score (RS) is the mean retained TPR across transformed conditions; worst-case is the minimum TPR among the named transformation pipelines.

Detector Clean TPR RS Worst-case pipeline TPR FAR after human editing
Fast-DetectGPT 97.7% 0.727 46.7% 3.4%
Binoculars-lite 96.9% 0.634 44.0% 7.4%
DeBERTa-v3, HC3 fine-tune 93.1% 0.829 78.7% 8.0%
GPT-2 perplexity 93.1% 0.642 62.0% 6.4%
OpenAI RoBERTa detector 85.9% 0.554 53.3% 6.7%
TF-IDF + logistic regression 79.4% 0.961 63.3% 1.1%
Stylometric GBM 55.7% 0.915 50.7% 1.2%

What the current analysis suggests:

  • Strong clean performance does not guarantee transformation robustness. The two strongest clean detectors fall to 44–47% TPR under the depth-2 laundering pipeline.
  • The supervised DeBERTa model retains the highest worst-case pipeline TPR, while TF-IDF has the highest relative robustness at a lower clean TPR.
  • False-accusation rates vary substantially by detector, transformation, and writing register. The slice results do not support a simple “native versus non-native” explanation.
  • Quality-adjusted evasion results are provisional until the semantic-similarity cutoff has been checked against human judgments.

Full generated tables are available in PER_DOMAIN.md, REPORT.md, FAIRNESS.md, HARDNESS.md, and STATS.md.

Evaluation design

  • Symmetric transformations: apply each transformation to human and machine text.
  • Clean-only training: trainable detectors do not see transformed text during training.
  • Frozen thresholds: select a target operating point once and do not retune it for each transformation.
  • Document-level grouping: keep a source document and all of its variants in the same split.
  • Reliability metrics: report RS, worst-case performance, false-accusation rate, calibration drift, quality-adjusted evasion, and hardness collapse.
  • Cluster-aware inference: resample source documents rather than individual rows and correct paired comparisons for multiple testing.

The current draft contains both pooled-threshold and per-domain sensitivity analyses. A separate held-out calibration set is required before the reported operating-point estimates should be treated as final.

Quickstart

The lightweight installation runs the metric, statistics, split, detector, and transformation tests without downloading model weights:

git clone https://github.com/razon1494/stress-test.git
cd stress-test
python -m pip install -e ".[dev]"
python -m pytest -q

For the complete pipeline:

python -m pip install -e ".[models,data,dev]"
python scripts/01_build_dataset.py --help
python scripts/02_apply_transforms.py --help
python scripts/03_run_detectors.py --help
python scripts/04_make_report.py --help

Some fairness corpora require separate downloads and cannot be redistributed in this repository. See the reproducibility guide before attempting a full run.

Repository map

Path Purpose
src/stress_test/ Transformations, detectors, metrics, statistics, data utilities, and report generation
scripts/ Numbered experiment and analysis stages
tests/ Unit tests for core research logic
results/ Committed generated summaries and structured outputs
paper/ LaTeX manuscript draft and generated figure
docs/ Reproduction instructions, annotation protocol, and related-work notes

Current limitations

  • The machine-text core currently uses one generator family and one broad source dataset (HC3).
  • Threshold calibration and final evaluation are not yet separated in the committed result set.
  • Semantic preservation is currently embedding-scored; the planned three-annotator validation has not been completed.
  • The DeBERTa fine-tune uses one training seed, and Binoculars-lite is not the original Falcon-7B model pair.
  • Several subgroup estimates have wide uncertainty because the cells are small.
  • Raw licensed corpora, model weights, and score caches are intentionally not committed.

Citation

Until a preprint is released, cite the software repository:

@software{rahman2026stresstest,
  author = {Rahman, Mohammad Arifur},
  title  = {STRESS-Test: Semantic-preserving Transformations for Robust
            Evaluation of Synthetic-text Screening},
  year   = {2026},
  url    = {https://github.com/razon1494/stress-test}
}

Author and license

Mohammad Arifur Rahman · rahman.arif.cse@gmail.com · Portfolio

The framework code is released under the MIT License. External datasets and pretrained models remain subject to their original licenses and terms of use.

About

Reliability evaluation framework for AI-text detectors — hardness-aware, semantics-controlled, symmetric false-accusation analysis under real-world transformations.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages