Pith. sign in

REVIEW 4 major objections 6 minor 44 references

By 2026, four of ten U.S. and Chinese government document streams that scored near zero in 2021 show statistically significant traces of AI-assisted writing, with the U.S. signal downstream of policy work and the PRC signal closer to it.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-11 17:37 UTC pith:V67OOCMP

load-bearing objection Clean pilot that turns public-document AI detection into a usable governance monitoring signal; the four-source 2026 result is real on the detector they used, but Chinese calibration remains the soft underbelly. the 4 major comments →

arxiv 2607.04543 v1 pith:V67OOCMP submitted 2026-07-05 cs.CY

Government AI Use as a Monitoring Primitive: A Public Document Pilot Study

David I. Atkinson , Joan Eleanor O'Bryan This is my paper
classification cs.CY
keywords government AI adoptionAI text detectionecosystem monitoringpublic documentsrevealed behaviorU.S.–Chinafrontier AI governancemonitoring primitive
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Governments matter for frontier AI governance, yet their real day-to-day use of AI is hard to see: procurement and official statements lag, select, and track formal adoption more than practice. This paper proposes a cheap complementary monitor—measuring language-model writing traces in the public documents governments already publish—and treats the resulting score as a standardized monitoring primitive based on revealed behavior. In a pilot of more than 3,000 documents from ten U.S. and PRC streams, 2021 baselines sit near zero; by 2026 four sources rise with confidence intervals excluding zero. In the sample the U.S. signal concentrates in Military Review and DARPA news (downstream of high-level policy), while the PRC signal concentrates at CAC and MOST (closer to rule-writing and science-policy administration). A sympathetic reader cares because the method is externally reproducible by journalists, academics, or other states, can flag activity missing from incomplete self-reported inventories, and can be composed with other instruments without requiring insider access.

Core claim

Across ten public U.S. and PRC government-related document streams, mean per-document AI-writing fractions are near zero in 2021. By 2026 four sources show statistically significant elevation: U.S. Military Review (0.10) and DARPA news (0.08) sit downstream of policy generation, while PRC CAC (0.13) and MOST (0.11) sit closer to it; country-pooled means reach roughly 0.05–0.07 with bootstrap confidence intervals excluding zero. The paper presents this as a pilot demonstration that public-document AI detection can function as a lightweight monitoring primitive for government AI use.

What carries the argument

The monitoring primitive is the document-level AI fraction ai in [0,1] returned by a commercial AI-text classifier (Pangram) that segments each public document into windows, classifies each window, and averages them; the same score is applied to any fixed panel of public streams so that changes over time and across sources can be tracked and bootstrapped.

Load-bearing premise

The central claim depends on the detector’s per-document AI fraction being a sufficiently accurate measure of real language-model assistance in these English and Chinese government genres, rather than an artifact of formal bureaucratic style, domain shift, or language-specific detector bias.

What would settle it

Have independent expert human annotators score a large stratified sample of the same 2021 and 2024–2026 documents for AI assistance; if the human-judged rise is absent or far smaller than the detector rise, or if the detector still flags high fractions on matched known-human bureaucratic text, the claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Journalists, academics, NGOs, or other states can re-run the same panel monthly on public outputs without insider access or government self-report.
  • Revealed-behavior scores can surface AI activity omitted from incomplete agency use-case inventories and from DoD/intelligence exemptions.
  • Where the signal sits (downstream of policy vs. near policy formulation) can indicate which parts of the state adopt first.
  • Score changes can prioritize sources for qualitative review, procurement searches, interviews, or comparison with disclosed AI policies.
  • As model-family attribution matures, the same streams may indicate domestic versus foreign or closed versus open model use.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If policy-near PRC sources continue to lead, export controls may constrain commercial more than state adoption, especially where governments retain privileged model access.
  • Absence of signal in high-level U.S. policy outlets may understate elite exposure if AI is used for analysis but not for the final public text.
  • Once the primitive is public, deliberate detector evasion by agencies becomes itself a measurable governance response.
  • Expanding the panel across more languages, agencies, and non-AI-keyworded streams would test whether the U.S.–PRC reverse pattern generalizes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes measuring traces of language-model assistance in public government documents as a lightweight, externally reproducible monitoring primitive for government AI adoption. In a pilot of ten U.S. and PRC government-related document streams (n≈3,068), it reports near-zero Pangram AI fractions in 2021 and rising scores from 2024, with four sources showing statistically significant 2026 elevations (CAC, MOST, PLA Daily, Military Review). It further claims that, in this sample, the U.S. signal concentrates in outlets downstream of policy work while the PRC signal concentrates closer to policy formulation, and discusses how the signal could complement procurement and disclosure instruments while noting detector brittleness and limited Chinese validation.

Significance. If the detector scores are valid measures of LLM assistance in these genres, the paper supplies a useful, low-cost ecosystem-monitoring instrument for technical AI governance: revealed-behavior evidence that is cheap to recompute, comparable across jurisdictions, and less dependent on government self-report than use-case inventories or procurement records. Strengths include a clean pre-LLM 2021 baseline near zero across all sources, transparent per-source counts and multi-level bootstrap CIs (Tables 1–2, Figures 1–4), carefully scoped pilot claims rather than causal claims about high-level policymakers, and an explicit limitations discussion. The framing as a composable monitoring primitive is a genuine contribution even if the numerical country pattern remains provisional.

major comments (4)
  1. [§1 Method; §2 Limitations; Table 2] §1 Method and §2 Limitations, together with Table 2: the headline claim that four of ten sources show significant 2026 AI-assisted writing, and the U.S.-downstream vs PRC-policy-proximate contrast, rest primarily on Pangram fraction_ai for three Chinese sources (CAC 0.13 [0.07,0.19], MOST 0.11 [0.02,0.21], PLA Daily 0.14 [0.09,0.21]). The manuscript itself states that independent Chinese-language validation is limited. A detector that is well-calibrated on 2021 pre-LLM text can still systematically over-flag formal Chinese bureaucratic or military-commentary style (or under-flag English policy prose) once post-2023 fluency conventions appear. Without a targeted validation—e.g., human expert annotation of a stratified Chinese subsample, dual-detector comparison, or a held-out Chinese bureaucratic calibration set—the four-source count and the country-pattern interpretation remain vulnerabl
  2. [Table 1; Table 2; Figures 3–4] Table 1 and the 2026 columns of Table 2 / Figures 3–4: several 2026 cells are very small or partial (OSTP n=5, DARPA n=13, State Council n=19, MOST n=42; cutoff 2026-04-25). Wide CIs for DARPA (0.08 [0.00,0.20]) and MOST (0.11 [0.02,0.21]) mean that “statistically significant signs” and the ranking that underpins the downstream-vs-proximate narrative are sensitive to a handful of documents. The paper should either restrict the 2026 significance claim to sources with adequate n, report a pre-registered minimum cell size, or show leave-one-out / influence diagnostics so readers can see whether a few high-scoring items drive the means.
  3. [Abstract; §1; Data availability] Abstract and §1 (“externally reproducible”) vs. data availability: the primitive is marketed as lightweight and externally reproducible from public outputs, yet the manuscript does not release document IDs, scraped text, or Pangram window-level scores. Without that release, independent re-scoring with alternative detectors (or future Pangram versions) is impossible, which undercuts both the reproducibility claim and the ability of civil-society actors—the intended users—to verify or extend the panel. Releasing at least URLs, hashes, and per-document fraction_ai (even if full text is restricted) would make the contribution match its framing.
  4. [§1 Results; Abstract] §1 results paragraph on U.S. “downstream of high-level policy generation” vs PRC “closer to” policy: this contrast is interpretive and post-hoc. Military Review and DARPA news are labeled downstream; OSTP and AI-keyworded Federal Register are labeled policy-proximate and show no signal; CAC and MOST are labeled policy-proximate and do show signal. The operational definition of “downstream” vs “closer to policy work” is not pre-specified, and alternative groupings (e.g., military vs civilian, commentary vs normative instruments) could reorganize the pattern. Either pre-register a proximity coding, report a sensitivity table under alternative codings, or demote the contrast from a main empirical claim to a suggestive observation.
minor comments (6)
  1. [Figure 1] Figure 1 pools countries with equal source weights; a sentence in the caption or §1 stating how sensitive the pooled means are to dropping PLA Daily or Military Review would help readers assess robustness of the country-level rise.
  2. [Section C] Section C source descriptions are thorough, but a short justification for excluding other natural candidates (e.g., DoD press, State Department, NPC documents) would clarify selection bias risk for the pilot panel.
  3. [Section B] Section B quality check is valuable; reporting the post-remediation exclusion count in the main text (not only appendix) would reassure readers that placeholder/stub remediation does not drive the 2026 elevations.
  4. [Table 2] In Table 2, bolding intervals that exclude zero is helpful; consider also marking which of the four “significant” sources survive a multiple-comparison correction if the paper continues to count them as a set.
  5. [Section A.3] Related work §A.3 could briefly note how the government panel differs from Liang et al. peer-review/scientific-paper estimates in genre formality, which may affect detector false-positive rates.
  6. [AI Usage Statement] The AI Usage Statement is appropriate and clear; keep it.

Circularity Check

0 steps flagged

Empirical measurement of detector scores on public corpora; no circular derivation or fitted-as-prediction steps.

full rationale

The paper's central claims are observational: collect public documents from ten sources, run a commercial AI-text classifier (Pangram) to obtain per-document fraction_ai in [0,1], then report means and bootstrap CIs by source and year, with 2021 as a pre-LLM baseline. The four-source significance finding and the U.S./PRC location contrast are direct numerical outputs of that pipeline (Table 1 counts, Table 2 means/CIs, Figures 1–4). There is no parameter fitted to a subset of the same data and then re-labeled a prediction; no equation that defines the target quantity in terms of itself; no uniqueness theorem or ansatz imported from the authors' prior work that forces the result; and no renaming of a known pattern as a new derivation. Related-work citations (including one co-authored paper on algorithmic progress) are background only and do not enter the measurement chain. Detector validity is a correctness/assumption risk, not circularity. The derivation is therefore self-contained against external benchmarks and exhibits no circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 1 invented entities

The central claim rests on measurement assumptions rather than free parameters or new physical entities. Load-bearing premises are domain assumptions about detector validity, baseline validity, and the interpretive link between public-text AI fraction and government AI adoption. No numerical constants are fitted to produce the headline means; the detector itself is treated as a black-box instrument.

axioms (4)
  • domain assumption Pangram’s windowed AI-fraction scores are a sufficiently accurate proxy for language-model assistance on both English and Chinese government documents of the genres studied.
    Invoked as the measurement instrument throughout §1 and defended by citation to independent evaluations; paper itself notes limited independent Chinese validation.
  • domain assumption 2021 public documents constitute a valid near-zero pre-mass-market-LLM baseline for these sources.
    Used as the control year for all comparisons; empirically supported by the near-zero observed scores but still an interpretive premise.
  • domain assumption Elevated AI fractions in published government text are informative about government AI adoption/use relevant to AI governance, even if they primarily capture drafting or public-affairs writing.
    Core interpretive link stated in the introduction and Discussion; paper acknowledges the risk that the signal measures bureaucratic production more than high-level policy use.
  • ad hoc to paper Equal weighting of sources and equal weighting of documents within source is an appropriate aggregation for the pooled country means.
    Explicit design choice in Figure 1 caption; alternative weightings (by document volume or policy centrality) are not reported.
invented entities (1)
  • public-document AI-detection monitoring primitive no independent evidence
    purpose: To name a standardized, externally reproducible measurement of revealed government AI writing assistance that can be tracked over time and composed with other governance instruments.
    Framing contribution of the paper; not a physical entity but a methodological construct introduced to organize the pilot results.

pith-pipeline@v1.1.0-grok45 · 40487 in / 2890 out tokens · 37531 ms · 2026-07-11T17:37:49.962000+00:00 · methodology

0 comments
read the original abstract

Governments are important actors in frontier AI governance, but many facts about their adoption and use of AI systems are difficult to observe directly. Procurement disclosures and official statements are useful, but can also be delayed, selective, and better suited to measuring formal adoption than actual day-to-day use. We propose a complementary monitoring primitive: measuring traces of language-model assistance in public government documents. The approach is lightweight, externally reproducible, and based on revealed behavior rather than stated intent. In a pilot study of ten public document streams from U.S. and PRC government-related sources, we find that, while 2021 baselines are consistently near zero, by 2026, four of our ten sources show statistically significant signs of AI-assisted writing. In our sample, the U.S. signal concentrates in publications downstream of policy work; the PRC signal concentrates closer to it. We close by discussing how this signal could complement existing instruments for monitoring government AI adoption, and where it falls short.

Figures

Figures reproduced from arXiv: 2607.04543 by David I. Atkinson, Joan Eleanor O'Bryan.

Figure 1
Figure 1. Figure 1: Pooled AI-writing signal by country, 2021–2026. Each marker is the mean per-document fraction of text flagged as AI￾generated by Pangram (range 0–1), pooled across each country’s sources ( [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Mean fraction of each document flagged as AI-generated by Pangram, for each source across time. Shaded bands show 95% bootstrap confidence intervals. All sources show near-zero baselines in 2021; both countries exhibit rising scores from 2024 onward, with Chinese sources reaching a source-weighted pooled mean of 0.07 [95% CI: 0.02, 0.12] by 2026 and U.S. sources 0.05 [0.01, 0.11]. public document stream—th… view at source ↗
Figure 3
Figure 3. Figure 3: Per-source mean AI fraction over time for the four U.S. sources (Military Review, DARPA news, Federal Register, OSTP), with 95% bootstrap confidence intervals. All panels share a common y-axis. E. Random Examples For each source we show one randomly drawn document with fraction ai < 0.1 and up to 2 randomly drawn documents with fraction ai > 0.5 (both conditional such documents existing). Bodies are trunca… view at source ↗
Figure 4
Figure 4. Figure 4: Per-source mean AI fraction over time for the six PRC sources (PLA Daily, CAC, MOST, and the three gov.cn streams), with 95% bootstrap confidence intervals. All panels share the same y-axis scale as [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

44 extracted references · 4 canonical work pages

  1. [1]

    C., Tripto, N

    Ansari, A., Zhang, D. C., Tripto, N. I., and Lee, D. Echoes of Automation : The Increasing Use of LLMs in Newsmaking , April 2026

    2026

  2. [2]

    Claude Gov models for U.S

    Anthropic . Claude Gov models for U.S. national security customers. Anthropic News, June 2025. URL https://www.anthropic.com/news/claude-gov-models-for-u-s-national-security-customers

    2025

  3. [3]

    From text to source: Results in detecting large language model-generated content

    Antoun, W., Sagot, B., and Seddah, D. From text to source: Results in detecting large language model-generated content. In Calzolari, N., Kan, M.-Y., Hoste, V., Lenci, A., Sakti, S., and Xue, N. (eds.), Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pp.\ 7531--7543...

    2024

  4. [4]

    Anthropic Economic Index report: Uneven geographic and enterprise AI adoption, November 2025

    Appel, R., McCrory, P., Tamkin, A., McCain, M., Neylon, T., and Stern, M. Anthropic Economic Index report: Uneven geographic and enterprise AI adoption, November 2025

    2025

  5. [5]

    Military review: The professional journal of the U.S

    Army University Press . Military review: The professional journal of the U.S. army. Army University Press, U.S. Army Combined Arms Center, Fort Leavenworth, KS, n.d. URL https://www.armyupress.army.mil/Military-Review/

  6. [6]

    Racing for recognition? Theorizing emerging status hierarchies and prestige competition in the AI era

    Blomquist, K. Racing for recognition? Theorizing emerging status hierarchies and prestige competition in the AI era. International Affairs, 102 0 (3): 0 949--970, 2026. doi:10.1093/ia/iiag028. URL https://doi.org/10.1093/ia/iiag028

    work page doi:10.1093/ia/iiag028 2026

  7. [7]

    and Jia, X

    Chen, C. and Jia, X. When researchers use AI : Public trust, ethical judgments, and the perceived value of academic research. AI and Ethics, 6 0 (2): 0 223, March 2026. ISSN 2730-5961. doi:10.1007/s43681-026-01039-w

    work page doi:10.1007/s43681-026-01039-w 2026

  8. [8]

    and Lau, K.-s

    Cheung, S. and Lau, K.-s. DeepSeek Use in PRC Military and Public Security Systems . China Brief 25(20), The Jamestown Foundation, October 2025. URL https://jamestown.org/deepseek-use-in-prc-military-and-public-security-systems/

    2025

  9. [9]

    Communications and public affairs

    Defense Advanced Research Projects Agency . Communications and public affairs. https://www.darpa.mil/news/public-affairs, n.d

  10. [10]

    Adding Clinical trial registry entries, Earnings calls prepared remarks and Federal court opinions from CourtListener

    Denain, J.-S. Adding Clinical trial registry entries, Earnings calls prepared remarks and Federal court opinions from CourtListener . X (formerly Twitter), Epoch AI, March 2026. URL https://x.com/i/status/2037646276251279674

    arXiv 2026

  11. [11]

    and Spero, M

    Emi, B. and Spero, M. Technical Report on the Pangram AI-Generated Text Classifier , July 2024

    2024

  12. [12]

    Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government

    Executive Office of the President . Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government . Executive Order 13960, Federal Register, document 2020-27065, December 2020. URL https://www.federalregister.gov/documents/2020/12/08/2020-27065/promoting-the-use-of-trustworthy-artificial-intelligence-in-the-federal-government. Publish...

    2020

  13. [13]

    The state of AI competition in advanced economies

    Haag, A. The state of AI competition in advanced economies. Feds notes, Board of Governors of the Federal Reserve System, Washington, D.C., October 2025. URL https://www.federalreserve.gov/econres/notes/feds-notes/the-state-of-ai-competition-in-advanced-economies-20251006.html

    2025

  14. [14]

    Spotting LLMs With Binoculars : Zero-Shot Detection of Machine-Generated Text , October 2024

    Hans, A., Schwarzschild, A., Cherepanova, V., Kazemi, H., Saha, A., Goldblum, M., Geiping, J., and Goldstein, T. Spotting LLMs With Binoculars : Zero-Shot Detection of Machine-Generated Text , October 2024

    2024

  15. [15]

    Rage against the machine? Generative AI exposure, subjective risk, and policy preferences

    Haslberger, M., Gingrich, J., and Bhatia, J. Rage against the machine? Generative AI exposure, subjective risk, and policy preferences. Journal of European Public Policy, September 2025. doi:10.1080/13501763.2025.2554903. Advance online publication, September 10, 2025

    work page doi:10.1080/13501763.2025.2554903 2025

  16. [16]

    C., Atkinson, D., Thompson, N., and Sevilla, J

    Ho, A., Besiroglu, T., Erdil, E., Owen, D., Rahman, R., Guo, Z. C., Atkinson, D., Thompson, N., and Sevilla, J. Algorithmic progress in language models, March 2024

    2024

  17. [17]

    and Imas, A

    Jabarian, B. and Imas, A. Artificial writing and automated detection. Working Paper 34223, National Bureau of Economic Research, 2025. URL https://www.nber.org/papers/w34223

    2025

  18. [18]

    Legacy procurement practices shape how u.s

    Johnson, N., Silva, E., Leon, H., Eslami, M., Schwanke, B., Dotan, R., and Heidari, H. Legacy procurement practices shape how u.s. cities govern ai: Understanding government employees’ practices, challenges, and needs. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’25, pp.\ 772–789. ACM, 2025. doi:10.1145/3...

    doi:10.1145/3715275.3732049 2025

  19. [19]

    Anthropic disables fable and mythos ai models after u.s

    Kahn, J. Anthropic disables fable and mythos ai models after u.s. government bars it from giving foreigners access. Fortune, June 2026. URL https://fortune.com/2026/06/13/anthropic-disables-fable-mythos-export-controls-national-security-threat/

    2026

  20. [20]

    Agencies report over 3,000 AI use cases in 2025

    Kelley, A. Agencies report over 3,000 AI use cases in 2025. Nextgov/FCW, April 2026. URL https://www.nextgov.com/artificial-intelligence/2026/04/agencies-report-over-3000-ai-use-cases-2025/412898/. Published April 16, 2026. Reports that the Intelligence Community and Department of Defense are exempt from inventory reporting

    2025

  21. [21]

    A Watermark for Large Language Models , May 2024

    Kirchenbauer, J., Geiping, J., Wen, Y., Katz, J., Miers, I., and Goldstein, T. A Watermark for Large Language Models , May 2024

    2024

  22. [22]

    GPT detectors are biased against non-native English writers, July 2023

    Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., and Zou, J. GPT detectors are biased against non-native English writers, July 2023

    2023

  23. [23]

    A., and Zou, J

    Liang, W., Izzo, Z., Zhang, Y., Lepp, H., Cao, H., Zhao, X., Chen, L., Ye, H., Liu, S., Huang, Z., McFarland, D. A., and Zou, J. Y. Monitoring AI-Modified Content at Scale : A Case Study on the Impact of ChatGPT on AI Conference Peer Reviews , March 2024

    2024

  24. [24]

    D., and Zou, J

    Liang, W., Zhang, Y., Wu, Z., Lepp, H., Ji, W., Zhao, X., Cao, H., Liu, S., He, S., Huang, Z., Yang, D., Potts, C., Manning, C. D., and Zou, J. Quantifying large language model usage in scientific papers. Nature Human Behaviour, 9 0 (12): 0 2599--2609, December 2025. ISSN 2397-3374. doi:10.1038/s41562-025-02273-8

    doi:10.1038/s41562-025-02273-8 2025

  25. [25]

    Feature-augmented transformers for robust ai-text detection across domains and generators, 2026

    Mady, M., Reschke, J., and Schuller, B. Feature-augmented transformers for robust ai-text detection across domains and generators, 2026. URL https://arxiv.org/abs/2605.03969

    Pith/arXiv arXiv 2026

  26. [26]

    Misra, A., Wang, J., McCullers, S., White, K., and Ferres, J. L. Measuring AI Diffusion : A Population-Normalized Metric for Tracking Global AI Usage , November 2025

    2025

  27. [27]

    D., and Finn, C

    Mitchell, E., Lee, Y., Khazatsky, A., Manning, C. D., and Finn, C. DetectGPT : Zero-Shot Machine-Generated Text Detection using Probability Curvature , July 2023

    2023

  28. [28]

    M., Xiao, S., Xu, Y., and Yang, Y

    Naughton, B., Cheung, T. M., Xiao, S., Xu, Y., and Yang, Y. Reorganization of China 's science and technology system. IGCC Working Paper, UC Institute on Global Conflict and Cooperation, 2023. URL https://ucigcc.org/publication/reorganization-of-chinas-science-and-technology-system/

    2023

  29. [29]

    Newhard, J. M. The stock market speaks: How Dr . Alchian learned to build the bomb. Journal of Corporate Finance, 27: 0 116--132, August 2014. ISSN 0929-1199. doi:10.1016/j.jcorpfin.2014.05.002

    work page doi:10.1016/j.jcorpfin.2014.05.002 2014

  30. [30]

    2025 federal agency artificial intelligence use case inventory

    Office of Management and Budget . 2025 federal agency artificial intelligence use case inventory. https://github.com/ombegov/2025-Federal-Agency-AI-Use-Case-Inventory, 2025. Compiled pursuant to E.O.\ 13960, the Advancing American AI Act, and OMB Memorandum M-25-21

    2025

  31. [31]

    S., Rajkumar, N., Mo \"e s, N., Ladish, J., Guha, N., Newman, J., Bengio, Y., South, T., Pentland, A., Koyejo, S., Kochenderfer, M

    Reuel, A., Bucknall, B., Casper, S., Fist, T., Soder, L., Aarne, O., Hammond, L., Ibrahim, L., Chan, A., Wills, P., Anderljung, M., Garfinkel, B., Heim, L., Trask, A., Mukobi, G., Schaeffer, R., Baker, M., Hooker, S., Solaiman, I., Luccioni, A. S., Rajkumar, N., Mo \"e s, N., Ladish, J., Guha, N., Newman, J., Bengio, Y., South, T., Pentland, A., Koyejo, S...

    2024

  32. [32]

    People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI -generated text

    Russell, J., Karpinska, M., and Iyyer, M. People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI -generated text. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp.\ 5342--5373, Vienna, Austria, 2025. Association for Computational Linguistics. URL htt...

    2025

  33. [33]

    S., Kumar, A., Balasubramanian, S., Wang, W., and Feizi, S

    Sadasivan, V. S., Kumar, A., Balasubramanian, S., Wang, W., and Feizi, S. Can ai-generated text be reliably detected?, 2025. URL https://arxiv.org/abs/2303.11156

    Pith/arXiv arXiv 2025

  34. [34]

    Compute Trends Across Three Eras of Machine Learning

    Sevilla, J., Heim, L., Ho, A., Besiroglu, T., Hobbhahn, M., and Villalobos, P. Compute Trends Across Three Eras of Machine Learning . In 2022 International Joint Conference on Neural Networks ( IJCNN ) , pp.\ 1--8, July 2022. doi:10.1109/IJCNN55064.2022.9891914

    doi:10.1109/ijcnn55064.2022.9891914 2022

  35. [35]

    and Levy, J

    Shah, A. and Levy, J. Access to Justice in the Age of AI : Evidence from U . S . Federal Courts . SSRN working paper, March 2026. URL https://ssrn.com/abstract=6766859

    2026

  36. [36]

    China's AI regulations and how they get made

    Sheehan, M. China's AI regulations and how they get made. Technical report, Carnegie Endowment for International Peace, July 2023. URL https://carnegieendowment.org/research/2023/07/chinas-ai-regulations-and-how-they-get-made

    2023

  37. [37]

    Sutter, K. M. U.s. export controls and china: Advanced semiconductors. CRS Report R48642, Congressional Research Service, September 2025. URL https://www.congress.gov/crs-product/R48642. Updated September 19, 2025

    2025

  38. [38]

    Federal Agencies Largely Miss the Mark on Documenting AI Compliance Plans as Required by AI Executive Order

    Tobin-Miyaji , M. Federal Agencies Largely Miss the Mark on Documenting AI Compliance Plans as Required by AI Executive Order . EPIC, November 2024. URL https://epic.org/federal-agencies-largely-miss-the-mark-on-documenting-ai-compliance-plans-as-required-by-ai-executive-order/. Published November 21, 2024

    2024

  39. [39]

    Department of Defense

    U.S. Department of Defense . Military and Security Developments Involving the People 's Republic of China . Technical report, December 2025

    2025

  40. [40]

    Government Accountability Office

    U.S. Government Accountability Office . Artificial Intelligence: Agencies Have Begun Implementation but Need to Complete Key Requirements . Technical Report GAO-24-105980, U.S. Government Accountability Office , December 2023. URL https://www.gao.gov/products/gao-24-105980

    2023

  41. [41]

    Government Accountability Office

    U.S. Government Accountability Office . Artificial intelligence: Generative ai use and management at federal agencies. Report to Congressional Requesters GAO-25-107653, U.S. Government Accountability Office, July 2025. URL https://www.gao.gov/assets/gao-25-107653.pdf

    2025

  42. [42]

    Congress on Superintelligence

    Wildeford, P. Congress on Superintelligence . The AI Policy Network (AIPN), June 2026. URL https://theaipn.org/issue/quotes/

    2026

  43. [43]

    S., and Wong, D

    Wu, J., Yang, S., Zhan, R., Yuan, Y., Chao, L. S., and Wong, D. F. A Survey on LLM-Generated Text Detection : Necessity , Methods , and Future Directions . Computational Linguistics, 51 0 (1): 0 275--338, March 2025. ISSN 0891-2017. doi:10.1162/coli_a_00549

    doi:10.1162/coli_a_00549 2025

  44. [44]

    and Mathur, V

    Zimmerman, M. and Mathur, V. Ombegov/2024- Federal-AI-Use-Case-Inventory . Office of the Federal Chief Information Officer, April 2026

    2024