Pith. sign in

REVIEW 3 major objections 4 minor 79 references

PRAC3 (Privacy, Reputation, Accountability, Consent, Credit, Compensation): Long Tailed Risks of Voice Actors in AI Data-Economy

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Voice actors face long-tailed risks that a C3-only ethics cannot capture, and this paper introduces the PRAC3 framework to address them.

desk verdict A worthwhile qualitative study of voice actor risk, but the paper's claim that PRAC3 substantively extends C3 is asserted more than proven; the empirical material justifies conditional publication. read the letter →

arxiv 2507.16247 v1 pith:AJTJX7XI submitted 2025-07-22 cs.CY cs.AIcs.HC

Tanusree Sharma , Yihao Zhou , Visar Berisha This is my paper
classification cs.CYcs.AIcs.HC
keywords voiceactorssyntheticAIdataeconomybiometricidentitylong-tailedrisksPRAC3frameworkgovernancegenerativeharms
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 tries to establish that professional voice actors are exposed to long-tailed risks—harms that emerge long after a recording is made and through uses the actor never agreed to—because a voice is simultaneously creative work and a biometric identifier. It argues that the standard ethical frame of Consent, Credit, and Compensation (C3) cannot capture these harms, which include reputational damage from decontextualized reuse, identity theft and fraud using cloned voices, and accountability gaps when no one can be traced or held responsible. Using interviews with 20 professional US voice actors, the paper documents concrete incidents in each category and presents the PRAC3 framework—Privacy, Reputation, Accountability, Consent, Credit, Compensation—as an interdependent set of risk dimensions for the synthetic voice economy. A sympathetic reader would care because if the framework is right, governance, contracts, and technical protections for voice data need to be redesigned around more than payment and credit.

What carries the argument

The load-bearing mechanism of the paper is the PRAC3 framework, a six-dimensional risk model for voice data in the AI economy: Privacy (unauthorized exposure of biometric identity), Reputation (harm from decontextualized or misaligned reuse), Accountability (legal and technical gaps in traceability and recourse), plus the inherited Consent, Credit, and Compensation. The framework is built as a threat model: for each risk scenario it identifies the asset at stake (the voiceprint, the professional persona, or contractual rights), the threat actor (client, platform, third-party modder, cybercriminal), the vulnerability, and the potential impact. It is carried by the interview data coded thematically, and by the four actor personas, which explain why the same risks land differently on newcomers without representation than on established actors with agents, lawyers, or union support. The framework does the work of converting scattered, post-hoc incident reports into a forward-looking checklist for anticipating low-probability, high-impact harms.

What would settle it

Conduct a comparable interview or survey study with voice actors in two or three non-US regions and ask them to map their experienced harms onto the six PRAC3 dimensions; if a substantial share of harms falls outside the six categories, or if actors in those regions do not recognize the accountability gap as the binding constraint, then the framework's cross-context claim is falsified. A narrower check: if robust, widely deployed voice-provenance tools existed and were in routine use by actors, the paper's claim that traceability mechanisms are absent would be empirically contradicted.

Watch

Extended reading notes

Core claim

The paper's central discovery is that voice is a double asset—expressive labour and a stable biometric voiceprint—and that this duality creates a distinct class of harms that existing ethical frameworks miss. As voice recordings circulate through audition platforms, client contracts, and public datasets, they can be cloned, recontextualized, and redeployed without enforceable constraints, so the original actor faces fallout that is social, financial, and legal at once. The paper reports that actors find their voices in ads they never recorded, in AI-generated adult content, in political messages they do not endorse, and in clones used for fraud, while having no mechanism to trace, contest, or remove the misuse. From these experiences, the paper derives the PRAC3 framework, which adds Privacy, Reputation, and Accountability to the familiar C3 pillars of Consent, Credit, and Compensation, and it organizes actors into four personas—Emerging Professional, Solo Defender, Delegator, Strategist—to show how resources and experience shape exposure. The claim, stated on the paper's own terms, is that PRAC3 captures the context-transcending, long-tailed risks of synthetic voice replication and offers a conceptual basis for future governance.

Load-bearing premise

The framework's generalizability rests on the assumption that the experiences of 20 voice actors, all based in the United States, represent the risk landscape of voice actors broadly; if labor protections, legal regimes, or platform conditions differ elsewhere, the relative weight of the six pillars—and even which risks emerge—could change.

Editorial extensions

If this is right

  • If PRAC3 is right, voice data in AI training should be governed as biometric personal data, not merely as creative content, which changes the default consent and retention obligations.
  • Contractual protections would need to cover the full lifecycle—audition samples, delivered files, and downstream AI training—rather than just the initial performance, because harm can arise long after delivery.
  • Provenance and watermarking for voice become a governance requirement, since accountability cannot be assigned without traceability of how a voiceprint was obtained and reused.
  • Union and legal resources would be prioritized for the Emerging Professional and Solo Defender personas, who face the same risks with the least recourse.
  • Dataset builders and platforms that host voice work would need to anticipate that 'public' or 'voluntary' contributions can later be repurposed into commercial synthetic voices, and engineer opt-in and opt-out accordingly.

Reading between the lines

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

  • Editorial inference: the same double-asset logic likely extends beyond voice to other biometric-adjacent creative outputs—such as a person's face, gait, or distinctive drawing style—so the PRAC3 pillars may transfer to the synthetic video and image economy, a connection the paper does not make.
  • The paper leaves implicit a testable extension: a comparative interview study outside the United States could reveal whether the six pillars hold where union structures, right-of-publicity laws, and platform ecosystems differ; the authors flag the US-only sample as a limitation.
  • Editorial inference: a quantitative follow-up could build a risk-scoring instrument from the six dimensions and measure whether actors' self-assessed exposure tracks actual misuse incidents, turning the conceptual framework into an assessment tool.
  • Another editorial extension: the accountability pillar could be operationalized as a technical requirement for verifiable provenance in text-to-speech systems, not just a legal category.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This paper reports a qualitative interview study of 20 professional voice actors in the United States, examining how they perceive and respond to risks arising from generative AI voice replication. The authors describe risks across the workflow of discovery, auditioning, contracting, and recording/file-sharing, and they document incidents of unauthorized cloning, reputational harm, financial fraud, and accountability gaps. Based on these interviews, they introduce the PRAC3 framework, which expands the existing Consent, Credit, Compensation (C3) framing with three additional pillars: Privacy, Reputation, and Accountability. The paper argues that voice is both creative labor and a biometric identifier, and that long-tailed risks emerge when vocal data is decoupled from context, authorship, and control. The paper includes an ethics statement and positionality statement that acknowledge the US-only sample and the conceptual, non-validated status of the framework.

Significance. If the added pillars can be shown to be analytically distinct from C3, the paper would make a useful contribution by giving voice actors a grounded threat-modeling vocabulary and by documenting concrete, understudied harms in the synthetic voice economy. The study's strengths include the use of direct participant quotes, a transparent description of the interview and coding process, a tabulated set of incident examples, and honest disclosure of the sample's geographic limits. The paper does not attempt mathematical or machine-checked claims, but its qualitative evidence is plausibly reported. The main open question is whether the paper actually demonstrates that Privacy, Reputation, and Accountability add substantive analytical power beyond C3, or whether they are re-descriptions of consent, credit, and compensation failures; this question is load-bearing for the central claim and is not yet resolved.

major comments (3)
  1. [Discussion: Ethical Frameworks: From C³ to PRAC³; Table 2] The central claim that C3 does not adequately address emergent risks is not supported by a comparative analysis. In Table 2, almost every incident is mapped to both a new pillar and at least one C3 pillar: Incident 1 is labeled Consent, Compensation, Accountability; Incident 2 is Reputation, Consent, Accountability; Incident 3 is Consent, Compensation, Accountability. These mappings can be read as showing that the harms are already covered by consent, credit, and compensation, with the new pillars serving as consequences or mitigations rather than independent risk dimensions. To make the 'beyond C3' claim load-bearing, the authors should conduct a comparative coding exercise: apply a C3-only codebook to the interview excerpts and demonstrate which harms remain uncoded, or provide an analytic argument for why Privacy, Reputation, and Accountability are not logically implied by violations of consent, credit, and compensation. Without this, the contribution may reduce to re-labeling.
  2. [Method: Data Analysis] The paper's deductive code list includes Participant Category, Awareness and Understanding of AI Risks, Workflow and Practices, Ownership and Compensation, and Privacy and Security Concerns, but no a priori code for Reputation or Accountability. This makes it unclear whether the three new pillars were genuinely induced from the data or imposed by the researchers' prior framing. The method section also reports no codebook excerpt, no inter-coder agreement measure, and no procedure for resolving coding disagreements. Because the central claim is that PRAC3 is grounded in voice actors' lived experiences, the authors should provide a more complete audit trail: for example, include the final codebook with definitions, show representative quotes for each new pillar, and explain how the themes of Privacy, Reputation, and Accountability were derived from the coded transcripts.
  3. [Table 2; Figure 2] The framework's own illustrative mapping is internally inconsistent. Incidents 4 and 12 in Table 2 list 'Identity' as part of the PRAC³ Domain, but Identity is not one of the six pillars and is not defined in Figure 2. Additionally, Figure 2 defines Accountability as 'lack of legal or technical resources to trace, attribute, or address misuse of voice data,' which is a governance or recourse gap rather than a risk dimension parallel to Privacy and Reputation. These inconsistencies suggest post-hoc labeling and weaken the table's evidentiary value. The authors should either define Identity as a sub-component of Privacy, align all table labels with the six pillars, and clarify whether Accountability is a distinct harm dimension or a failure of mitigation and recourse.
minor comments (4)
  1. [Results: Accountability and Legal Uncertainty] The participant identifier 'P117' appears in the text and should be 'P17'; similarly, the phrase 'mismassed' in the Reputational and Ethical Risks subsection appears to be a typo for 'misused' or 'misappropriated.'
  2. [Table 1; Results: Personas of Voice Actors] The textual description of the Delegator persona states 'more than 5+ years of experience,' but Table 1 categorizes the Delegator as 'Low experience, High resources.' These statements are contradictory and should be reconciled.
  3. [Ethics Statement; Discussion] The paper honestly discloses that all interviewees are from the United States, but the Discussion and Conclusion are written in general terms about 'governance models' and 'AI Data ecosystems.' Since the framework is derived from a US-only sample, the authors should consistently scope their claims as applying to the US context until further cross-cultural validation is available.
  4. [Abstract and Introduction] The abstract states that LibriSpeech was built with 'hundreds of individual contributors,' but LibriSpeech is derived from LibriVox audiobooks and involves many more than a few hundred narrators. Please verify the number or phrase it as 'thousands of volunteer narrators.'

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: PRAC3 is presented as an inductively derived framework grounded in interview data, not as a mathematically forced prediction or a self-citation chain.

full rationale

This paper is a qualitative interview study, so the kinds of circularity that arise from fitted parameters, definitional identities, or imported uniqueness theorems do not apply. The central claim is that professional voice actors face risks beyond consent, credit, and compensation, and that the PRAC3 framework captures these risks. The framework is explicitly derived from the interview data: the Ethics Statement says 'The PRAC3 framework proposed in this paper is derived mainly from the information provided to us by the respondents in the interviews,' and the Method section describes a deductive-inductive thematic analysis. This is transparent inductive synthesis, not a derivation whose conclusion is equivalent to its input. The skeptical concern that the new pillars (Privacy, Reputation, Accountability) may be redundant with C3 is a substantive contribution or validity question, but it is not circularity: the paper does not define the new pillars in terms of C3, and Table 2's co-occurrence of both types of labels in incident mappings does not show that the new pillars were constructed from C3. The authors do cite their own prior work (e.g., Sharma et al. 2023a, 2023b, 2024, 2025; Yu et al. 2024; Kaushik et al. 2024), but those citations support background claims about data-sharing practices and privacy concerns; they are not load-bearing for the PRAC3 framework's derivation. The acknowledged geographic limitation about U.S.-only interviewees is a sampling and generalizability limitation, not a circular step. Overall, there is no circular step that reduces the paper's central result to its own inputs.

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

The paper is a qualitative interview study, so it has no free parameters in the usual sense. Its main load-bearing assumptions are the representativeness and interpretation of the interview data, plus the assumed biometric persistence of voice identity. The PRAC3 framework is an invented conceptual entity, but it is presented as a framing tool rather than a new empirical observation.

assumptions (3)
  • domain assumption The 20 interviewed US-based voice actors are sufficiently representative of voice actors in the synthetic voice economy.
    The paper generalizes from a sample of 20 US-based voice actors to a framework for voice actors broadly. The Ethics Statement acknowledges all interviewees are from the US, which limits geographic generalizability.
  • domain assumption Thematic analysis of interview transcripts accurately surfaces the participants' risk perceptions.
    The paper relies on the coding and thematic analysis process described in the Method section, but the codebook and inter-rater reliability are not fully reported, so the analysis is an interpretive step.
  • domain assumption A person's voice is a biometric identifier that remains identifiable even after transformation or anonymization.
    The paper states that the underlying biometric voice signature often remains partially detectable by machines even if data is anonymized, but does not provide experimental evidence for this claim, instead relying on prior biometric literature.
invented entities (1)
  • PRAC3 framework
    purpose: A conceptual framework to assess and categorize long-tailed risks to voice actors in the AI data economy, expanding C3 by adding Privacy, Reputation, and Accountability.
    PRAC3 is a conceptual classification scheme derived from the authors' interview data. It does not make a falsifiable prediction or provide an independent testable handle outside the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PRAC3 (Privacy, Reputation, Accountability, Consent, Credit, Compensation): Long Tailed Risks of Voice Actors in AI Data-Economy." pith.science (2026). https://pith.science/paper/AJTJX7XI

@misc{pith2026250716247,
  author       = {Pith},
  title        = {Pith review of: PRAC3 (Privacy, Reputation, Accountability, Consent, Credit, Compensation): Long Tailed Risks of Voice Actors in AI Data-Economy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AJTJX7XI}},
  note         = {Machine review of arXiv:2507.16247}
}
read the original abstract

Early large-scale audio datasets, such as LibriSpeech, were built with hundreds of individual contributors whose voices were instrumental in the development of speech technologies, including audiobooks and voice assistants. Yet, a decade later, these same contributions have exposed voice actors to a range of risks. While existing ethical frameworks emphasize Consent, Credit, and Compensation (C3), they do not adequately address the emergent risks involving vocal identities that are increasingly decoupled from context, authorship, and control. Drawing on qualitative interviews with 20 professional voice actors, this paper reveals how the synthetic replication of voice without enforceable constraints exposes individuals to a range of threats. Beyond reputational harm, such as re-purposing voice data in erotic content, offensive political messaging, and meme culture, we document concerns about accountability breakdowns when their voice is leveraged to clone voices that are deployed in high-stakes scenarios such as financial fraud, misinformation campaigns, or impersonation scams. In such cases, actors face social and legal fallout without recourse, while very few of them have a legal representative or union protection. To make sense of these shifting dynamics, we introduce the PRAC3 framework, an expansion of C3 that foregrounds Privacy, Reputation, Accountability, Consent, Credit, and Compensation as interdependent pillars of data used in the synthetic voice economy. This framework captures how privacy risks are amplified through non-consensual training, how reputational harm arises from decontextualized deployment, and how accountability can be reimagined AI Data ecosystems. We argue that voice, as both a biometric identifier and creative labor, demands governance models that restore creator agency, ensure traceability, and establish enforceable boundaries for ethical reuse.

Figures

Figures reproduced from arXiv: 2507.16247 by the authors.

Figure 1
Figure 1. Risks and AI-related threats in different stages of voice acting work, including discovery, audition, contracting, [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. The PRAC³ framework, including six risk dimensions in the use of voice actors’ data in the context of generative AI: [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

79 extracted references · 67 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    A3MLAnticipatory. 2025. A3ML: Anticipatory and Adaptive Anti-Money Laundering. https://www.darpa.mil/research/programs/a3ml-anticipatory-adaptive. [Accessed 23-05-2025]

    work page 2025

  4. [4]

    K.; and Loewenstein, G

    Acquisti, A.; John, L. K.; and Loewenstein, G. 2013. What is privacy worth? The Journal of Legal Studies, 42(2): 249--274

    work page 2013

  5. [5]

    Act, E. A. 2023. E U A I A ct: first regulation on artificial intelligence | T opics | E uropean P arliament --- europarl.europa.eu. https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence. [Accessed 23-05-2025]

    arXiv 2023

  6. [6]

    AESDD. 2018. Acted Emotional Speech Dynamic Database – AESDD. https://m3c.web.auth.gr/research/aesdd-speech-emotion-recognition/. [Accessed 27-01-2025]

    work page 2018

  7. [7]

    H.; Awumey, E.; and Das, S

    Agnew, W.; Barnett, J.; Chu, A.; Hong, R.; Feffer, M.; Netzorg, R.; Jiang, H. H.; Awumey, E.; and Das, S. 2024. Sound Check: Auditing Audio Datasets. arXiv preprint arXiv:2410.13114

    arXiv 2024

  8. [8]

    S.; and Katsaggelos, A

    Aleksic, P. S.; and Katsaggelos, A. K. 2006. Audio-visual biometrics. Proceedings of the IEEE, 94(11): 2025--2044

    work page 2006

Show all 79 references
  1. [9]

    Allen, A. L. 1999. Privacy-as-data control: Conceptual, practical, and moral limits of the paradigm. Conn. L. Rev., 32: 861

    1999

  2. [10]

    Allyn, B. 2023. ' N ew Y ork T imes' sues C hat G P T creator O pen A I , M icrosoft, for copyright infringement. https://www.npr.org/2023/12/27/1221821750/new-york-times-sues-chatgpt-openai-microsoft-for-copyright-infringement. [Accessed 22-05-2025]

    2023

  3. [11]

    Andalibi, N.; Ozturk, P.; and Forte, A. 2017. Sensitive self-disclosures, responses, and social support on Instagram: The case of\# depression. In Proceedings of the 2017 ACM conference on computer supported cooperative work and social computing, 1485--1500

    2017

  4. [12]

    M.; and Weber, G

    Ardila, R.; Branson, M.; Davis, K.; Henretty, M.; Kohler, M.; Meyer, J.; Morais, R.; Saunders, L.; Tyers, F. M.; and Weber, G. 2019. Common voice: A massively-multilingual speech corpus. arXiv preprint arXiv:1912.06670

    2019 arXiv

  5. [13]

    Baroni, M.; Bernardini, S.; Ferraresi, A.; and Zanchetta, E. 2009. The WaCky wide web: a collection of very large linguistically processed web-crawled corpora. Language resources and evaluation, 43: 209--226

    2009

  6. [14]

    Bateman, J. 2020. Deepfakes and Synthetic Media in the Financial System: Assessing Threat Scenarios. Technical report, Carnegie Endowment for International Peace

    2020

  7. [15]

    Baumgartner, J.; Zannettou, S.; Keegan, B.; Squire, M.; and Blackburn, J. 2020. The pushshift reddit dataset. In Proceedings of the international AAAI conference on web and social media, volume 14, 830--839

    2020

  8. [16]

    BBC. 2021. Actor sues TikTok for using her voice in viral tool. https://www.bbc.com/news/technology-57063087. [Accessed 23-05-2025]

    2021

  9. [17]

    Blaising, A.; and Dabbish, L. 2022. Managing the transition to online freelance platforms: self-directed socialization. Proceedings of the ACM on Human-Computer Interaction, 6(CSCW2): 1--26

    2022

  10. [18]

    Bloomberg. 2024. youtube says openai training sora with its video would break rules. https://www.bloomberg.com/news/articles/2024-04-04/youtube-says-openai-training-sora-with-its-videos-would-break-the-rules?sref=10lNAhZ9&embedded-checkout=true. [Accessed 22-05-2025]

    2024

  11. [19]

    CCM, C. 2024. CSA Cloud Control Matrix (CCM v4). [Accessed 27-01-2025]

    2024

  12. [20]

    Chakrabarty, T.; Padmakumar, V.; Brahman, F.; and Muresan, S. 2024. Creativity Support in the Age of Large Language Models: An Empirical Study Involving Professional Writers. In Proceedings of the 16th Conference on Creativity & Cognition, 132--155

    2024

  13. [21]

    Cho, W. 2024. YouTube Creators Step Into Legal Battle Against OpenAI With Class Action Lawsuit. https://www.hollywoodreporter.com/business/business-news/youtube-creators-step-legal-battle-against-openai-class-action-lawsuit-1235968822/. [Accessed 23-05-2025]

    2024

  14. [22]

    Clarke, V.; and Braun, V. 2017. Thematic analysis. The journal of positive psychology, 12(3): 297--298

    2017

  15. [23]

    Code; and Culture. 2024. The 3 C’s of Voice Acting in the Age of AI: Consent, Control & Compensation. https://cultureandcode.io/gilfry-interview/. [Accessed 23-05-2025]

    2024

  16. [24]

    Communications, I. 2024. I A T S E welcomes release of bipartisan U . S . S enate roadmap for A I policy, urges C ongressional action - I A T S E --- iatse.net. https://iatse.net/iatse-welcomes-release-of-bipartisan-u-s-senate-roadmap-for-ai-policy-urges-congressional-action/#...

    2024

  17. [25]

    Cook, A. 2024. Illinois BIPA: A Litigation Nightmare for Employers. UIC Law Review, 57(2): 5

    2024

  18. [26]

    L., Ade Adetunji

    Ekene Chuks-Okeke, B. L., Ade Adetunji. 2023. Voice actors and generative AI: Legal challenges and emerging protection. https://iapp.org/news/a/voice-actors-and-generative-ai-legal-challenges-and-emerging-protections. [Accessed 23-05-2025]

    2023

  19. [27]

    EU. 2015. Creating Value through Open Data. https://data.europa.eu/sites/default/files/edp_creating_value_through_open_data_0.pdf. [Accessed 22-05-2025]

    2015

  20. [28]

    Gao, H.; Zahedi, M.; Treude, C.; Rosenstock, S.; and Cheong, M. 2024. Documenting ethical considerations in open source ai models. In Proceedings of the 18th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement, 177--188

    2024

  21. [29]

    GDPR. 2024. GDPRPrivacy Impact Assessment. [Accessed 27-01-2025]

    2024

  22. [30]

    I.; Desai, M.; Schnitzler, C.; Eom, N.; Cushman, J.; and Glassman, E

    Gero, K. I.; Desai, M.; Schnitzler, C.; Eom, N.; Cushman, J.; and Glassman, E. L. 2025. Creative Writers' Attitudes on Writing as Training Data for Large Language Models. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 1--16

    2025

  23. [31]

    Godard, B.; Schmidtke, J.; Cassiman, J.-J.; and Aym \'e , S. 2003. Data storage and DNA banking for biomedical research: informed consent, confidentiality, quality issues, ownership, return of benefits. A professional perspective. European journal of human genetics, 11(2): S88--S122

    2003

  24. [32]

    L.; and Suri, S

    Gray, M. L.; and Suri, S. 2019. Ghost work: How to stop Silicon Valley from building a new global underclass. Harper Business

    2019

  25. [33]

    Hoffman, S. 2015. Citizen science: the law and ethics of public access to medical big data. Berkeley Tech. LJ, 30: 1741

    2015

  26. [34]

    Hutiri, W.; Papakyriakopoulos, O.; and Xiang, A. 2024. Not my voice! a taxonomy of ethical and safety harms of speech generators. In Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, 359--376

    2024

  27. [35]

    Jasserand, C. 2024. Deceptive Deepfakes: Is the Law Coping with AI-Altered Representations of Ourselves? In 2024 International Conference of the Biometrics Special Interest Group (BIOSIG), 1--4. IEEE

    2024

  28. [36]

    Kagan, O. 2020. CCPA Regulations: Are Audio Recordings Personal Information? https://dataprivacy.foxrothschild.com/2020/06/articles/california-consumer-privacy-act/ccpa-regulations-audio-recordings/. [Accessed 23-05-2025]

    2020

  29. [37]

    Kang, D.; Hashimoto, T.; Stoica, I.; and Sun, Y. 2022. Zk-img: Attested images via zero-knowledge proofs to fight disinformation. arXiv preprint arXiv:2211.04775

    2022 arXiv

  30. [38]

    Kastrenakes, J. 2021. TikTok settles lawsuit with actress over its original text-to-speech voice. https://www.theverge.com/2021/9/29/22701167/bev-standing-tiktok-lawsuit-settles-text-to-speech-voice. [Accessed 22-05-2025]

    2021

  31. [39]

    F.; Wang, Y.; and Zou, Y

    Kaushik, S.; Sharma, T.; Yu, Y.; Ali, A. F.; Wang, Y.; and Zou, Y. 2024. Cross-Country Examination of People’s Experience with Targeted Advertising on Social Media. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems, 1--10

    2024

  32. [40]

    Kearns, J. 2014. Librivox: Free public domain audiobooks. Reference Reviews, 28(1): 7--8

    2014

  33. [41]

    S.; Binns, R.; Zhao, J.; and Biega, A

    Kyi, L.; Mahuli, A.; Silberman, M. S.; Binns, R.; Zhao, J.; and Biega, A. J. 2025. Governance of Generative AI in Creative Work: Consent, Credit, Compensation, and Beyond. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 1--16

    2025

  34. [42]

    Lane, J.; Stodden, V.; Bender, S.; and Nissenbaum, H. 2014. Privacy, big data, and the public good: Frameworks for engagement. Cambridge University Press

    2014

  35. [43]

    Lee, J.; and Hong, I. B. 2016. Predicting positive user responses to social media advertising: The roles of emotional appeal, informativeness, and creativity. International Journal of Information Management, 36(3): 360--373

    2016

  36. [44]

    Lefkovitz, N.; and Boeckl, K. 2020. NIST Privacy Framework: An Overview

    2020

  37. [45]

    Liang, C.; Peng, J.; Li, Z.; and Yin, M. 2024. The valuation paradox of generative AI: Evidence from gig workers. Available at SSRN 4825716

    2024

  38. [46]

    License, G. G. P. 1989. Gnu general public license. Retrieved December, 25: 2014

    1989

  39. [47]

    Lin, T.-Y.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Doll \'a r, P.; and Zitnick, C. L. 2014. Microsoft coco: Common objects in context. In European conference on computer vision, 740--755. Springer

    2014

  40. [48]

    Liu, Y.; Chen, S.; Cheng, H.; Yu, M.; Ran, X.; Mo, A.; Tang, Y.; and Huang, Y. 2024. How ai processing delays foster creativity: Exploring research question co-creation with an llm-based agent. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems, 1--25

    2024

  41. [49]

    T.; Cutrell, E.; Morrison, C.; Hofmann, K.; and Stumpf, S

    Massiceti, D.; Zintgraf, L.; Bronskill, J.; Theodorou, L.; Harris, M. T.; Cutrell, E.; Morrison, C.; Hofmann, K.; and Stumpf, S. 2021. Orbit: A real-world few-shot dataset for teachable object recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vis...

    2021

  42. [50]

    Miller, P.; Styles, R.; and Heath, T. 2008. Open data commons, a license for open data. LDOW, 369

    2008

  43. [51]

    Nagano, Y. 2025. California Creatives Rally Behind State AI Rules to Save Their Artwork. https://www.sfpublicpress.org/california-creatives-rally-behind-state-ai-rules-to-save-their-artwork. [Accessed 23-05-2025]

    2025

  44. [52]

    Nissenbaum, H. 2004. Privacy as contextual integrity. Wash. L. Rev., 79: 119

    2004

  45. [53]

    on Health Sciences Policy, B.; and on Strategies for Responsible Sharing of Clinical Trial Data, C. 2015. Sharing clinical trial data: maximizing benefits, minimizing risk

    2015

  46. [54]

    OpenAI. 2025. How ChatGPT and our foundation models are developed. https://help.openai.com/en/articles/7842364-how-chatgpt-and-our-foundation-models-are-developed. [Accessed 23-05-2025]

    2025

  47. [55]

    Panayotov, V.; Chen, G.; Povey, D.; and Khudanpur, S. 2015. Librispeech: an asr corpus based on public domain audio books. In 2015 IEEE international conference on acoustics, speech and signal processing (ICASSP), 5206--5210. IEEE

    2015

  48. [56]

    H.; Gurbani, V

    Pantiukhov, P.; Koriakov, D.; Petrova, T.; Alves, J. H.; Gurbani, V. K.; and State, R. 2024. Enhanced DeFi Security on XRPL with Zero-Knowledge Proofs and Speaker Verification. In 2024 IEEE International Conference and Expo on Real Time Communications at IIT (RTC), 23--30. IEEE

    2024

  49. [57]

    Prahallad, K. 2010. Automatic building of synthetic voices from audio books. Carnegie Mellon University

    2010

  50. [58]

    V.; and Black, A

    Prahallad, K.; Raghavendra, E. V.; and Black, A. W. 2010 a . Learning speaker-specific phrase breaks for text-to-speech systems. In SSW, 162--166

    2010

  51. [59]

    V.; and Black, A

    Prahallad, K.; Raghavendra, E. V.; and Black, A. W. 2010 b . Semi-supervised learning of acoustic driven prosodic phrase breaks for text-to-speech systems. In Proceedings of 5th International Conference on Speech Prosody (Speech Prosody 2010), Chicago, Illinois

    2010

  52. [60]

    Purdy, G. 2010. ISO 31000: 2009—setting a new standard for risk management. Risk Analysis: An International Journal, 30(6): 881--886

    2010

  53. [61]

    Romanosky, S.; and Acquisti, A. 2009. Privacy costs and personal data protection: Economic and legal perspectives. Berkeley Tech. LJ, 24: 1061

    2009

  54. [62]

    Russakovsky, O.; Deng, J.; Su, H.; Krause, J.; Satheesh, S.; Ma, S.; Huang, Z.; Karpathy, A.; Khosla, A.; Bernstein, M.; et al. 2015. Imagenet large scale visual recognition challenge. International journal of computer vision, 115(3): 211--252

    2015

  55. [63]

    Privacy is not for me, it's for those rich women

    Sambasivan, N.; Checkley, G.; Batool, A.; Ahmed, N.; Nemer, D.; Gayt \'a n-Lugo, L. S.; Matthews, T.; Consolvo, S.; and Churchill, E. 2018. " Privacy is not for me, it's for those rich women": Performative Privacy Practices on Mobile Phones by Women in South Asia. In Fourteent...

    2018

  56. [64]

    Schuhmann, C.; Beaumont, R.; Vencu, R.; Gordon, C.; Wightman, R.; Cherti, M.; Coombes, T.; Katta, A.; Mullis, C.; Wortsman, M.; et al. 2022. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural Information Processing Systems,...

    2022

  57. [65]

    Shan, S.; Cryan, J.; Wenger, E.; Zheng, H.; Hanocka, R.; and Zhao, B. Y. 2023. Glaze: Protecting artists from style mimicry by \ Text-to-Image \ models. In 32nd USENIX Security Symposium (USENIX Security 23), 2187--2204

    2023

  58. [66]

    Sharma, T. 2024. Inclusive. AI: Towards Democratic AI with DAO-Enabled Inclusive Decision-Making. OpenAI Grant Interim Report

    2024

  59. [67]

    I.; and Wang, Y

    Sharma, T.; Kaushik, S.; Yu, Y.; Ahmed, S. I.; and Wang, Y. 2023 a . User perceptions and experiences of targeted ads on social media platforms: Learning from bangladesh and india. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1--15

    2023

  60. [68]

    I'm not convinced that they don't collect more than is necessary

    Sharma, T.; Kyi, L.; Wang, Y.; and Biega, A. J. 2024. " I'm not convinced that they don't collect more than is necessary": \ User-Controlled \ Data Minimization Design in Search Engines. In 33rd USENIX Security Symposium (USENIX Security 24), 2797--2812

    2024

  61. [69]

    Sharma, T.; Potter, Y.; Kilhoffer, Z.; Huang, Y.; Song, D.; and Wang, Y. 2025. Aligning AI with Public Values: Deliberation and Decision-Making for Governing Multimodal LLMs in Political Video Analysis. arXiv preprint

    2025

  62. [70]

    Sharma, T.; Stangl, A.; Zhang, L.; Tseng, Y.-Y.; Xu, I.; Findlater, L.; Gurari, D.; and Wang, Y. 2023 b . Disability-first design and creation of a dataset showing private visual information collected with people who are blind. In Proceedings of the 2023 CHI Conference on Huma...

    2023

  63. [71]

    Tauberer, J. I. 2010. Learning [voice]. University of Pennsylvania

    2010

  64. [72]

    Tenbarge, K. 2024. Scarlett Johansson says she was 'shocked, angered' when she heard OpenAI's voice that sounded like her. https://www.nbcnews.com/tech/tech-news/scarlett-johansson-shocked-angered-openai-voice-rcna153180. [Accessed 23-05-2025]

    2024

  65. [73]

    A.; Friedland, G.; Elizalde, B.; Ni, K.; Poland, D.; Borth, D.; and Li, L.-J

    Thomee, B.; Shamma, D. A.; Friedland, G.; Elizalde, B.; Ni, K.; Poland, D.; Borth, D.; and Li, L.-J. 2016. Yfcc100m: The new data in multimedia research. Communications of the ACM, 59(2): 64--73

    2016

  66. [74]

    Tseng, Y.-Y.; Sharma, T.; Zhang, L.; Stangl, A.; Findlater, L.; Wang, Y.; Tseng, D. G. Y.-Y.; and Gurari, D. 2024. BIV-Priv-Seg: Locating Private Content in Images Taken by People With Visual Impairments. arXiv preprint arXiv:2407.18243

    2024 arXiv

  67. [75]

    Van Horn, R. 2007. Online books and audiobooks. Phi Delta Kappan, 89(2): 154

    2007

  68. [76]

    Yu, Y.; Sharma, T.; Hu, M.; Wang, J.; and Wang, Y. 2024. Exploring Parent-Child Perceptions on Safety in Generative AI: Concerns, Mitigation Strategies, and Design Implications. arXiv preprint arXiv:2406.10461

    2024 arXiv

  69. [77]

    Zarochintcev, S. 2021. Anticipatory Governance And National Security Risk Assessment. https://ideas.repec.org/a/nos/vgmu00/2021i3p200-218.html. [Accessed 23-05-2025]

    2021

  70. [78]

    W.; and Lee, M

    Zhang, A.; Boltz, A.; Wang, C. W.; and Lee, M. K. 2022. Algorithmic management reimagined for workers and by workers: Centering worker well-being in gig work. In Proceedings of the 2022 CHI conference on human factors in computing systems, 1--20

    2022

  71. [79]

    Zuboff, S. 2023. The age of surveillance capitalism. In Social theory re-wired, 203--213. Routledge

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.