Promising Topics for U.S.-China Dialogues on AI Risks and Governance

Aris Richardson; Conor McGlynn; Jeffrey Ding; Kristy Loke; Lujain Ibrahim; Marianne Lu; Saad Siddiqui; Stephen Clare

REVIEW 3 major objections 6 minor 51 references

Promising Topics for U.S.-China Dialogues on AI Risks and Governance

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper argues that a systematic comparison of 44 primary U.S. and Chinese AI governance documents reveals strong or moderate overlap in most major risk and governance categories, giving concrete topics for bilateral dialogue.

desk verdict A transparent, original-language comparison of US and Chinese AI governance documents that delivers a plausible map of common ground, with a real but non-fatal weakness in corpus representativeness. read the letter →

arxiv 2505.07468 v1 pith:6QJZSS7N submitted 2025-05-12 cs.CY

Saad Siddiqui , Lujain Ibrahim , Kristy Loke , Stephen Clare , Marianne Lu , Aris Richardson , Conor McGlynn , Jeffrey Ding This is my paper · ORCID

classification cs.CY

keywords AIgovernanceUS-Chinarelationsriskperceptionapproachesdocumentanalysisbilateraldialoguepolicy

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper asks whether the United States and China, despite strategic rivalry, have enough shared concerns about AI to make bilateral governance dialogue worthwhile. It answers by reading 44 primary government and corporate documents from both countries in their original languages, coding them for how they describe AI risks and what governance tools they favor. The central claim is that risk perceptions converge strongly on limited user transparency and poor reliability, and governance approaches converge strongly on using AI for safety and on multi-stakeholder convening. A broader set of areas—bias, robustness, dangerous capabilities, weak cybersecurity, external auditing, licensing or registration, watermarking, model evaluations, and adversarial testing—shows moderate overlap. If this mapping is right, the two governments have concrete, usable topics for talks even where their overall regulatory philosophies differ.

What carries the argument

The load-bearing instrument is an adapted coding taxonomy that classifies each passage of a governance document into named risk categories and named governance approaches, together with a three-level grading rule that labels cross-national agreement strong, moderate, or weak. The taxonomy's definitions determine what counts as a match: for example, transparency is compared by whether both sides require disclosure of how a system works to users, and adversarial testing is compared by whether red-team obligations exist and who bears them. This machinery lets the authors convert qualitative documents into comparable positions and then see where the two countries land in the same cell.

What would settle it

A direct check would be to re-run the same coding on a balanced post-2025 corpus: equal numbers of U.S. and Chinese government documents, current U.S. policy statements in place of the rescinded order, and updated Chinese standards. If the strong overlaps shrink to moderate or the moderate list changes substantially, the original classification was sample-driven; if the same categories persist, the finding is stable.

Watch

Extended reading notes

Core claim

The discovery is empirical: a systematic side-by-side reading of U.S. and Chinese AI policy texts finds more common ground than the conflict narrative suggests. On risk perception, the authors classify transparency and reliability as strong overlap, meaning U.S. and Chinese sources understand the problem in similar terms; they classify robustness, bias, interpretability, dangerous capabilities, and cybersecurity as moderate overlap, with real agreement but differing frames—for example, Chinese documents often treat chemical and biological risks under 'content security' while U.S. documents talk about capability proliferation. On governance, both sides strongly endorse using AI for safety and convening stakeholders; they moderately converge on external auditing, licensing and registration, watermarking, model evaluations, and adversarial testing. Privacy is the one risk category judged weakly aligned. The authors take these overlaps as evidence that dialogue topics can be selected pragmatically rather than waiting for agreement on first principles.

Load-bearing premise

The claim rests on the assumption that the chosen 44 documents—about twice as many Chinese as U.S. texts, with the U.S. side heavily represented by a 2023 executive order that was later rescinded—reflect each country's actual governance positions closely enough that the measured overlaps are not artifacts of the sample.

Editorial extensions

If this is right

Dialogues can open with the strongly aligned topics—transparency, reliability, AI-for-safety, and multi-stakeholder convening—where documents on both sides already speak in compatible terms.
Technical standards-setting talks could make early progress on commercial product safety, where reliability, robustness, and adversarial testing overlap, without touching national-security topics.
Content watermarking and provenance coordination is a concrete candidate for shared industry engagement, since both governments and leading firms support labeling mechanisms.
Model evaluations and red-teaming are a middle ground: both sides require them, but talks would need to bridge the U.S. focus on dangerous capabilities with China's focus on content security.
Privacy is the weak spot in risk perception, so bilateral dialogue should not begin there if the goal is quick common ground.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A balanced re-test with equal document counts per country and with current U.S. policy texts replacing the rescinded executive order would likely preserve some overlaps but could downgrade several moderate categories; the strong categories are the most sample-sensitive because they rest heavily on that single U.S. document.
The overlap on 'dangerous capabilities' may be shallower than the moderate label suggests: Chinese sources largely frame the issue as prohibited content while U.S. sources frame it as capability access, so agreeing on red lines would require resolving that frame difference first.
A testable extension would be to code post-2025 U.S. executive actions and China's updated generative-AI safety standard with the same taxonomy and check whether the strong-overlap set shrinks, which would locate how much of the finding was tied to the 2023–2024 document window.
Because companies on both sides read each other's technical reports, evaluation practices such as dangerous-capability testing could spread across borders even if formal regulation diverges, making corporate documents a leading indicator for future government alignment.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Worth a read if you work on AI governance or US-China relations. The paper does something genuinely useful: it codes 44 primary documents in the original languages and maps where US and Chinese actors converge on risk perception and governance approaches, using an adapted AGORA taxonomy. That granularity is new relative to earlier comparative work, and the authors are careful to distinguish strong, moderate, and weak overlap rather than forcing a binary. The quotes are documented, the coding framework is in the appendix, and the limitations section is honest about the fast-moving policy landscape.

The main empirical claim—that there are several areas of strong and moderate overlap—holds up better than the stress-test note suggests. The two 'strong' risk categories, limited user transparency and poor reliability, rest on NIST, Chinese standards, and MIIT, not just the Biden EO. The governance overlaps (pro-safety use of AI, convening) do lean more heavily on the EO, but they also have support from Tencent, Zhipu, CAICT, and Chinese standards. So the rescission of the EO would weaken but not collapse the headline result.

The real soft spots are two. First, the corpus asymmetry (30 Chinese vs 14 US documents) and the heavy weight on the Biden EO mean that the overlap grades could shift with a different document set. The authors acknowledge this but do not test robustness—a sensitivity analysis dropping the EO or substituting the January 2025 Trump order would have been straightforward and would have made the claims much stronger. Second, the coding is qualitative with no reported inter-coder reliability; the authors describe a collaborative review process but do not quantify agreement. That is a limitation, not a fatal flaw, for a scoping study of this kind.

There is also a minor issue: the paper was written before the EO rescission and the initial Note is a bit awkward, but Section 2.2 and Section 7 deal with it directly. The citation pattern looks clean; the self-citations (Ding's translations) support access, not the central claim.

Who is this for? AI governance researchers, diplomats, and people running Track II processes. It is a solid scoping study, not a definitive measurement. I would send it to peer review—it already is peer-reviewed, but if this were a new submission I would not desk-reject it. The claims are appropriately qualified and the method is transparent. My recommendation: engage with it, but ask for a sensitivity analysis and clearer coding reliability if it goes through another round.

Referee Report

3 major / 6 minor

Summary. The paper analyzes 44 primary U.S. and Chinese AI policy and corporate documents using an adapted version of the AGORA taxonomy, coding quoted spans for sociotechnical risks and governance approaches, then grading the degree of U.S.–China overlap as strong, moderate, or weak. It finds strong overlap on limited user transparency, poor reliability, the pro-safety role of AI, and stakeholder convening, and moderate overlap on robustness, bias, interpretability, dangerous capabilities, cybersecurity, external auditing, licensing/registration, watermarking, model evaluations, and adversarial testing. Based on these grades, the paper recommends specific topics for bilateral dialogue, including national-security-relevant evaluation standards, technical standardization, content provenance, and Track II discussions. The paper explicitly acknowledges that the analysis was completed before the rescission of the Biden AI Executive Order, that the corpus is asymmetric (about twice as many Chinese as U.S. documents), and that the coding is qualitative without formal inter-coder reliability metrics.

Significance. If the overlap findings are robust, the paper is a valuable and timely contribution: it provides systematic, original-language evidence for concrete common ground in U.S.–China AI governance, an area where claims of convergence or divergence are often made but rarely documented at the level of specific policy provisions. The use of fluent researchers for Chinese-language documents, the grounding in the AGORA taxonomy, and the explicit separation of risk perception from governance approaches are notable strengths. The paper is also appropriately cautious, noting that overlap does not imply endorsement or guaranteed dialogue value. However, the significance is conditional on the representativeness of the corpus, particularly the heavy reliance on the now-rescinded Biden AI EO, and on the reliability of the qualitative overlap grading. These issues are fixable but currently leave the headline result more fragile than the authors' framing suggests.

major comments (3)

[§3.4, Table 2; §7] The U.S. side of the corpus is dominated by the rescinded Biden AI EO: 161 of 189 U.S. quotes come from executive-branch documents, with the EO the largest single source. The paper acknowledges this in the initial Note, §2.2, and §7, but it never tests how the overlap grades in Table 5 change if the EO is removed or if post-rescission U.S. documents (e.g., the January 2025 executive order cited as [51]) are substituted. Because several categories graded 'strong' or 'moderate'—notably convening, the pro-safety role of AI, licensing/registration, external auditing, and watermarking—are illustrated primarily with EO provisions, the headline result could be an artifact of a single now-invalid document. I request a robustness analysis: re-grade the overlap categories excluding the EO and any other rescinded provisions, and report which grades persist, weaken, or disappear. This is essential for the paper's central claim that these topics are promising for ongoing dialogue.
[§3.2, §3.5, Table 4] The strong/moderate/weak classification criteria in Table 4 are defined qualitatively ('most, if not all' for strong; 'some overlap... key differences' for moderate), and the paper does not operationalize the thresholds or report inter-coder reliability metrics. The text states that two authors independently categorized each case of overlap and reached consensus, but no agreement statistics or per-category evidence counts are given. Without a table showing, for each category, how many U.S. and Chinese documents/quotes were coded under that category, the reader cannot assess whether a grade of 'strong' rests on broad cross-document support or on a small number of quotes. Please provide a supplemental evidence table (or an appendix) with quote counts per category per country, and clarify the decision rules used to map the presence of quotes to the three overlap grades.
[§3.1, §3.4] The corpus asymmetry (30 Chinese vs. 14 U.S. documents) and the differing sampling strategies—broad inclusion of Chinese corporate whitepapers versus focused sampling of model cards from three U.S. frontier labs—may bias the comparison toward convergence. A narrower U.S. set will tend to find fewer divergent positions, making 'overlap' easier to achieve. The paper acknowledges the numerical asymmetry but does not test how sensitive the Table 5 results are to alternative document selections, such as adding more U.S. corporate governance documents or excluding Chinese state/party documents that have no U.S. counterpart. At minimum, the authors should discuss whether the current design under- or over-estimates overlap, and ideally should run a sensitivity check with a more balanced corpus to show that the strong and moderate grades are not an artifact of selection.

minor comments (6)

[§5.2.3] The phrase 'appears oto be different' should read 'appears to be different'.
[§5.2.6] The phrase 'cloud compute U.S.ge' should read 'cloud compute usage'.
[§5.2.6] The FLOPS thresholds are rendered as '1026' and '1023'; they should be typeset as 10^26 and 10^23 with proper superscripts.
[Table 5 vs §5.1.1] Table 5 lists 'Pro-safety role of AI systems' while the section heading is 'Use of AI for pro-safety purposes'; harmonize the terminology across the paper.
[§3.5] The text says the analysis is restricted to 'areas of high and moderate overlap', but the grading scheme in Table 4 uses 'strong' rather than 'high'; use consistent labels.
[Appendix A] The acronym list includes UNCLOS and EEZ, which are not used in the body of the paper; either remove them or explain their relevance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's overlap findings are empirical codings of external primary documents, not derived from the authors' prior results or from fitted inputs.

full rationale

The paper's central claim—that several areas of strong and moderate overlap exist between U.S. and Chinese risk perceptions and governance approaches—is a qualitative empirical finding produced by applying an external taxonomy (AGORA, adapted with the authors' additions) to 44 primary documents and grading overlap according to explicit criteria in Table 4. No equation, fitted parameter, or prior result from the authors is used to construct the finding. The only self-references are Jeffrey Ding's translations of Chinese documents and the ChinAI newsletter, plus Ding's historical article on nuclear safety cooperation; none of these is load-bearing for the overlap grades. The taxonomy definitions and overlap rubric are stated independently of the outcome, and the coding process is described as line-by-line analysis with review and consensus, not as a derivation from the authors' prior conclusions. The paper's own limitations—the rescission of the Biden AI EO, corpus asymmetry, and the fact that identifying common ground does not establish dialogue value—are acknowledged in the Note, Section 2.2, and Section 7, but these are external-validity caveats, not circularity. No self-definitional, fitted-input, uniqueness-imported, ansatz-smuggling, or renaming pattern is present. The derivation chain runs from documents to codes to overlap grades to recommendations, and each step is evidenced by quoted document content rather than by construction from the conclusions. Therefore the score is 0.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities. It adds two categories to the AGORA taxonomy (dangerous capabilities and use of AI for pro-safety purposes), but these are classification dimensions, not invented entities. The main load-bearing choices are the classification thresholds and the assumptions about document representativeness and taxonomy validity.

free parameters (1)

Overlap classification thresholds (strong/moderate/weak)
The criteria in Table 4 are qualitative and were applied by the researchers to classify overlap. Different thresholds or more formal measures could change which areas are listed as strong or moderate overlap, affecting the recommendations.

assumptions (3)

domain assumption The AGORA taxonomy, adapted with two additional categories, is a valid and comprehensive framework for categorizing AI risks and governance approaches.
The paper relies on this taxonomy without independent validation; it is cited via a tool URL from CSET (Section 3.2).
domain assumption Primary policy and corporate documents accurately reflect the AI governance priorities and practices of their issuing entities.
The analysis treats documents as representative of national positions (Section 3.1), but implementation and unpublished context may differ; the paper itself notes this in Section 7.
domain assumption The selected documents are representative of U.S. and Chinese AI governance positions.
This is the weakest assumption; the corpus is asymmetric (roughly 2:1 Chinese to U.S. documents) and time-limited to documents collected in April 2024, with the U.S. side heavily weighted toward the now-rescinded Biden AI EO (Section 3.4 and the initial Note).

how reviews work

0 comments

Cite this review

Pith. "Pith review of Promising Topics for U.S.-China Dialogues on AI Risks and Governance." pith.science (2026). https://pith.science/paper/6QJZSS7N

@misc{pith2026250507468,
  author       = {Pith},
  title        = {Pith review of: Promising Topics for U.S.-China Dialogues on AI Risks and Governance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6QJZSS7N}},
  note         = {Machine review of arXiv:2505.07468}
}

read the original abstract

Cooperation between the United States and China, the world's leading artificial intelligence (AI) powers, is crucial for effective global AI governance and responsible AI development. Although geopolitical tensions have emphasized areas of conflict, in this work, we identify potential common ground for productive dialogue by conducting a systematic analysis of more than 40 primary AI policy and corporate governance documents from both nations. Specifically, using an adapted version of the AI Governance and Regulatory Archive (AGORA) - a comprehensive repository of global AI governance documents - we analyze these materials in their original languages to identify areas of convergence in (1) sociotechnical risk perception and (2) governance approaches. We find strong and moderate overlap in several areas such as on concerns about algorithmic transparency, system reliability, agreement on the importance of inclusive multi-stakeholder engagement, and AI's role in enhancing safety. These findings suggest that despite strategic competition, there exist concrete opportunities for bilateral U.S.-China cooperation in the development of responsible AI. Thus, we present recommendations for furthering diplomatic dialogues that can facilitate such cooperation. Our analysis contributes to understanding how different international governance frameworks might be harmonized to promote global responsible AI development.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 50 canonical work pages

[51]

Removing barriers to American leadership in Artificial Intelligence, January 2025

The White House. Removing barriers to American leadership in Artificial Intelligence, January 2025

work page 2025
[1]

Executive order 14105: Addressing United States investments in Certain National Security Technologies and products in Countries of Concern.Federal Register 88, 154 (2023), 54867–54872

work page 2023
[2]

Executive order 14110: Safe, secure, and trustworthy development and use of Artificial Intelligence.Federal Register 88, 210 (2023)

work page 2023
[3]

White paper on the governance and use of Generative Artificial Intelligence

Alibaba AI Governance Research Center. White paper on the governance and use of Generative Artificial Intelligence. White paper, Alibaba AI Governance Research Center, 10 2023

work page 2023
[4]

Analytica, O.Uk ai summit will promote some global cooperation.Emerald Expert Briefings, oxan-db (2023)

work page 2023
[5]

The Claude 3 model family: Opus, Sonnet, Haiku

Anthropic. The Claude 3 model family: Opus, Sonnet, Haiku. Model card, Anthropic, 3 2024. [6]Atleson, M.Keep your AI claims in check, 2 2023

work page 2024
[7]

Technical report, Baichuan Inc., 9 2023

Baichuan Inc.Baichuan 2: Open large-scale language models. Technical report, Baichuan Inc., 9 2023

work page 2023
[8]

Bengio, Y., et al.Managing extreme AI risks amid rapid progress.Science 384 (2024), 842–845

work page 2024

Show all 51 references

[9]

R.Executive order on the safe, secure, and trustworthy development and use of artificial intelligence

Biden, J. R.Executive order on the safe, secure, and trustworthy development and use of artificial intelligence
[10]

Birch, J., and Özfirat, Ö.Key takeaways from the AI Seoul summit 2024, 5 2024

2024
[11]

Politico EU, February 2025

Bristow, T.Britain dances to JD Vance’s tune as it renames AI institute. Politico EU, February 2025

2025
[12]

Brockmann, K.Applying export controls to ai: Current coverage and potential future controls.Armament, Arms Control and Artificial Intelligence: The Janus- faced Nature of Machine Learning in the Military Realm(2022), 193–209

2022
[13]

Author- itative large model AI safety benchmark first round results officially released, 4

China Academy of Information and Communications Technology. Author- itative large model AI safety benchmark first round results officially released, 4
[14]

China Academy of Information and Communications Technology. Reg- istration is now open: The first batch of evaluations of intelligent applications based on large models has been officially launched, and the standard publicity Promising Topics for U.S.–China Dialogues on AI Ris...

2025
[16]

The model Artificial Intelligence law (MAIL) v.2.0 - multilingual version

Chinese Academy of Social Sciences. The model Artificial Intelligence law (MAIL) v.2.0 - multilingual version. Model law, Chinese Academy of Social Sciences, 4 2024

2024
[17]

Chun, J., Schroeder de Witt, C., and Elkins, K.Comparative global AI regulation: Policy perspectives from the EU, China, and the US.arXiv preprint arXiv:2410.21279(10 2024)

2024 arXiv
[18]

China’s AI safety evaluations ecosystem.AI Safety in China(9 2024)

Concordia. China’s AI safety evaluations ecosystem.AI Safety in China(9 2024)

2024
[19]

Provisions on the management of algorithmic recommendations in Internet information services

Cyberspace Administration of China. Provisions on the management of algorithmic recommendations in Internet information services. Regulation, Cyberspace Administration of China, 12 2021. Translated by China Law Translate, January 2022

2021
[20]

Provisions on the administration of deep synthesis Internet information services, 2022

Cyberspace Administration of China. Provisions on the administration of deep synthesis Internet information services, 2022

2022
[21]

Interim measures for the management of Generative Artificial Intelligence services, 7 2023

Cyberspace Administration of China. Interim measures for the management of Generative Artificial Intelligence services, 7 2023. [22]Davis, B.Back on track?The Wire China(1 2024)

2023
[23]

Ding, J.ChinAI #261: First results from CAICT’s AI safety benchmark.ChinAI Newsletter(4 2024)

2024
[24]

Ding, J.Keep your enemies safer: technical cooperation and transferring nuclear safety and security technologies.European Journal of International Relations (2024), 13540661241246622

2024
[25]

Feldgoise, J., Dohmen, H., and Love, B.A Growing Yard: The Biden Admin- istration’s China Export Controls Are Ensnaring CPUs. Tech. rep., Center for Security and Emerging Technology, August 2024. [26]Gemini Team Google. Gemini: A family of highly capable multimodal models

2024
[27]

Giardini, T., and Fritz, J.The anatomy of AI rules: A systematic comparison of AI rules across the globe. Tech. rep., Digital Policy Alert, 6 2024

2024
[28]

Hine, E.Governing Silicon Valley and Shenzhen: Assessing a new era of Artificial Intelligence governance in the US and China. Tech. rep., 8 2023

2023
[29]

Hine, E., and Floridi, L.Artificial intelligence with American values and Chinese characteristics: A comparative analysis of American and Chinese governmental AI policies. Tech. rep., 1 2022

2022
[30]

C.Artificial intelligence, international competition, and the balance of power.2018 22(2018)

Horowitz, M. C.Artificial intelligence, international competition, and the balance of power.2018 22(2018)

2018
[31]

G.United nations ai resolution: A significant global policy effort to harness the technology for sustainable development: Iheid fc, May 2024

institute, G. G.United nations ai resolution: A significant global policy effort to harness the technology for sustainable development: Iheid fc, May 2024

2024
[32]

Lai, C., and Spring, J.Software must be secure by design, and Artificial Intelli- gence is no exception, 8 2023

2023
[33]

Guidelines for the construction of a national comprehensive standardization system for the Arti- ficial Intelligence industry

Ministry of Industry and Information Technology. Guidelines for the construction of a national comprehensive standardization system for the Arti- ficial Intelligence industry. Guidelines, Ministry of Industry and Information Technology, 1 2024

2024
[34]

National Institute of Standards and Technology. U.s. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools. Tech. rep., National Institute of Standards and Technology, August 2019

2019
[35]

Artificial intelligence risk management framework (AI RMF 1.0)

National Institute of Standards and Technology. Artificial intelligence risk management framework (AI RMF 1.0). Technical report, U.S. Department of Commerce, 1 2023

2023
[36]

White paper on AI security evaluation

National Quality Supervision and Inspection Center for Speech and Image Recognition Products. White paper on AI security evaluation. White paper, National Quality Supervision and Inspection Center for Speech and Image Recognition Products, 2023

2023
[37]

White paper on AI security evaluation

National Quality Supervision and Inspection Center for Speech and Image Recognition Products, National Industrial Information Security Development Research Center, and Institute of Artificial Intelligence. White paper on AI security evaluation. White paper, National Quality Su...

2021
[38]

Artificial intelligence computing platform information security framework

National Technical Committee 260 on Cybersecurity of Standardiza- tion Administration of China. Artificial intelligence computing platform information security framework. Technical standard, 5 2023

2023
[39]

Basic safety requirements for Generative Artificial Intelligence services

National Technical Committee 260 on Cybersecurity of Standardization Administration of China. Basic safety requirements for Generative Artificial Intelligence services. Technical standard, 2 2024. Translated by the Center for Security and Emerging Technology, April 2024

2024
[40]

red teaming

National Technical Committee 260 on Cybersecurity of Standardization Administration of China. Standards regarding security requirements for automated decision-making based on personal information. Technical standard, 2024. FAccT ’25, June 23–26, 2025, Athens, Greece Siddiqui e...

2024
[42]

F., Araujo, R., Belfield, H., Cleeland, B., Estier, M., Futerman, G., Guest, O., et al.The future of inter- national scientific assessments of ai’s risks, 2024

Pouget, H., Dennis, C., Batemen, J., Trager, R. F., Araujo, R., Belfield, H., Cleeland, B., Estier, M., Futerman, G., Guest, O., et al.The future of inter- national scientific assessments of ai’s risks, 2024

2024
[43]

Reuters, November 2024

Renshaw, J., and Hunnicutt, T.Biden, Xi agree that humans, not AI, should control nuclear arms. Reuters, November 2024

2024
[44]

Renshaw, J., and Hunnicutt, T.Biden, Xi agree that humans, not AI, should control nuclear arms.Reuters(11 2024)

2024
[45]

Reuters, May 2025

Revill, J., Farge, E., and Brunnstrom, D.Trump hails China talks, says two sides negotiated ’total reset’ in Geneva. Reuters, May 2025. [46]SenseTime. AI governance whitepaper. White paper, SenseTime, 2022

2025
[47]

AI (and other) companies: Quietly changing your terms of service could be unfair or deceptive, 2 2024

Staff in the Office of Technology and The Division of Privacy and Iden- tity Protection. AI (and other) companies: Quietly changing your terms of service could be unfair or deceptive, 2 2024

2024
[48]

Information security technology – assessment specification for security of machine learning algorithms

Standardization Administration of China. Information security technology – assessment specification for security of machine learning algorithms. Technical standard, Standardization Administration of China
[49]

Ten- cent large model security and safety report

Tencent Research Institute, Tencent Zhuqe Lab, Tencent Hunyuan Model Team, Tsinghua Shenzhen International Graduate School, and Zhejiang University State Key Lab of Blockchain and Data Security. Ten- cent large model security and safety report. Technical report, Tencent Resear...

2024
[50]

FACT SHEET: Biden-Harris administration secures voluntary commitments from leading Artificial Intelligence companies to manage the risks posed by AI, 7 2023

The White House. FACT SHEET: Biden-Harris administration secures voluntary commitments from leading Artificial Intelligence companies to manage the risks posed by AI, 7 2023

2023
[52]

The bletchley declaration by countries attending the ai safety summit, 1-2 november 2023, 2023

United Kingdom Government. The bletchley declaration by countries attending the ai safety summit, 1-2 november 2023, 2023. Accessed: 2025-01-17

2023
[53]

Eqal Employment Opportunity Commission

U.S. Eqal Employment Opportunity Commission. Select issues: Assessing adverse impact in software, algorithms, and Artificial Intelligence used in em- ployment selection procedures under Title VII of the Civil Rights Act of 1964. Guidance Title VII, 29 CFR Part 1607, U.S. Equal...

1964
[55]

Promising Topics for U.S.–China Dialogues on AI Risks and Governance FAccT ’25, June 23–26, 2025, Athens, Greece [56]Zeng, A., et al.GLM-130B: An open bilingual pre-trained model

Wu, X.Technology, power, and uncontrolled great power strategic competition between china and the united states.China International Strategy Review 2, 1 (2020), 99–119. Promising Topics for U.S.–China Dialogues on AI Risks and Governance FAccT ’25, June 23–26, 2025, Athens, Gr...

2020
[57]

Zeng, Y., Klyman, K., Zhou, A., Y ang, Y., Pan, M., Jia, R., Song, D., Liang, P., and Li, B.AI risk categorization decoded (AIR 2024): From government regulations to corporate policies

2024
[58]

M.US, China agree to expand military talks, continue AI cooperation after Sullivan-Wang meet.The Straits Times(8 2024)

Zhang, L. M.US, China agree to expand military talks, continue AI cooperation after Sullivan-Wang meet.The Straits Times(8 2024). Received 20 January 2025; accepted 9 April 2025

2024
[2024]

Translated by Jeffrey Ding, April 2024

2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [51]

Removing barriers to American leadership in Artificial Intelligence, January 2025

The White House. Removing barriers to American leadership in Artificial Intelligence, January 2025

work page 2025

[2] [1]

Executive order 14105: Addressing United States investments in Certain National Security Technologies and products in Countries of Concern.Federal Register 88, 154 (2023), 54867–54872

work page 2023

[3] [2]

Executive order 14110: Safe, secure, and trustworthy development and use of Artificial Intelligence.Federal Register 88, 210 (2023)

work page 2023

[4] [3]

White paper on the governance and use of Generative Artificial Intelligence

Alibaba AI Governance Research Center. White paper on the governance and use of Generative Artificial Intelligence. White paper, Alibaba AI Governance Research Center, 10 2023

work page 2023

[5] [4]

Analytica, O.Uk ai summit will promote some global cooperation.Emerald Expert Briefings, oxan-db (2023)

work page 2023

[6] [5]

The Claude 3 model family: Opus, Sonnet, Haiku

Anthropic. The Claude 3 model family: Opus, Sonnet, Haiku. Model card, Anthropic, 3 2024. [6]Atleson, M.Keep your AI claims in check, 2 2023

work page 2024

[7] [7]

Technical report, Baichuan Inc., 9 2023

Baichuan Inc.Baichuan 2: Open large-scale language models. Technical report, Baichuan Inc., 9 2023

work page 2023

[8] [8]

Bengio, Y., et al.Managing extreme AI risks amid rapid progress.Science 384 (2024), 842–845

work page 2024