Real Safety AI Foundation / Research

Privacy, Surveillance, and Consent

There Is No User in the Living Room: Ambient Household Devices and the User-Relative Architecture of the People-First Chatbot Act

Publisher: Real Safety AI Foundation

Research article.

Written
July 2026
Version
v0.4
Pages
35
Download PDFPlain language slides (PDF)Export BibTeXExport RIS

Abstract

The People-First Chatbot Act, introduced in the House on July 9, 2026, would impose privacy, transparency, and design duties on artificial intelligence chatbot providers. Every operative duty in the bill runs through a single term. A provider may process personal data only as necessary to fulfill an express request made by a user who has given affirmative consent. A provider must disable risky design features for a user it knows to be a minor. A provider incurs strict liability to a user injured through use. The Act then defines that term, at Section 5(22), as “an individual, regardless of age.” That is a genus without a differentia. It defines nothing. The defect is invisible on a phone, where the person holding the device and the person the statute regulates are the same person. It becomes dispositive on an ambient household device, a category already shipping into American homes at scale and about to be entered by the largest chatbot provider in the country. A child who walks past a camera-equipped speaker has made no request, consented to nothing, and interacts with the system only in the sense that the system is processing her face. This Article shows that the Act cannot say whether she is a user, that both available answers defeat the statute’s purpose, and that a third defect compounds them: the Act creates a private right of action denominated in violations while never defining a violation. The resulting damages exposure for a single afternoon in a single household ranges across four orders of magnitude on a question the text does not answer.

Keywords

  • People-First Chatbot Act
  • H.R. 9619
  • OpenAI
  • artificial intelligence
  • AI chatbots
  • ambient computing
  • smart speakers
  • children's privacy
  • COPPA
  • biometric data
  • age estimation
  • statutory construction
  • statutory damages
  • private right of action
  • bystander privacy

Plain language slides

First slide of the plain language summary of There Is No User in the Living Room: Ambient Household Devices and the User-Relative Architecture of the People-First Chatbot ActOpen the 26-slide summary (PDF)

Suggested citation

Gilly, Travis. "There Is No User in the Living Room: Ambient Household Devices and the User-Relative Architecture of the People-First Chatbot Act." Real Safety AI Foundation Research Article, July 2026. https://realsafetyai.org/research/na4a39/

Other versions

This paper is also posted on SSRN.

SSRN version

References (62)

This paper cites its sources in footnotes. Each authority is listed once, where it is first cited, with its footnote number.

  1. Footnote 2.M. Fairhurst, C. Li & M. Da Costa-Abreu, Predictive Biometrics: A Review and Analysis of Predicting Personal Characteristics from Biometric Data, 6 IET Biometrics 369 (2017)
  2. Footnote 2.E. Pantraki, C. Kotropoulos & A. Lanitis, Age Interval and Gender Prediction Using PARAFAC2 and SVMs Based on Visual and Aural Features, 6 IET Biometrics 290 (2017)
  3. Footnote 2.N. Lavan, The Effect of Familiarity on Within-Person Age Judgements from Voices, 113 Brit. J. Psychol. 287 (2021) (human listeners estimate age from voice with above-chance accuracy).
  4. Footnote 3.M.P. Ramaswamy & S. Palaniswamy, Multimodal Emotion Recognition: A Comprehensive Review, Trends, and Challenges, 14 WIREs Data Mining & Knowledge Discovery (2024)
  5. Footnote 4.H.R. 9619, 119th Cong. (2026) [hereinafter People-First Chatbot Act or the Act].
  6. Footnote 14.People-First Chatbot Act § 5(13).
  7. Footnote 17.People-First Chatbot Act, enacting clause
  8. Footnote 18.Press Release, Consumer Fed’n of Am., CFA Celebrates House Introduction of the People-First Chatbot Act (July 10, 2026), https://consumerfed.org/news/press-releases/cfa-celebrates-house-introduction-of-the-people-first-chatbot-act/.
  9. Footnote 20.Juli Clover, Jony Ive’s First OpenAI Device Will Be Smart Speaker With Camera, 2027 Launch Planned, MacRumors (Feb. 20, 2026), https://www.macrumors.com/2026/02/20/jony-ive-openai-smart-speaker-2027/ (last visited July 15, 2026).
  10. Footnote 22.People-First Chatbot Act § 2(a)(1)(A).
  11. Footnote 34.Open AI, Microsoft Sued Over ChatGPT’s Alleged Role in Fueling Man’s “Paranoid Delusions” Before Murder-Suicide in Connecticut, CBS News (CBS/AP) (Dec. 11, 2025), https://www.cbsnews.com/news/open-ai-microsoft-sued-chatgpt-murder-suicide-connecticut/ (last visited July 16, 2026).
  12. Footnote 36.15 U.S.C. §§ 6501–6506
  13. Footnote 36.16 C.F.R. pt. 312.
  14. Footnote 36.16 C.F.R. § 312.3.
  15. Footnote 37.People-First Chatbot Act § 5(22).
  16. Footnote 41.Gustafson v. Alloyd Co., 513 U.S. 561, 575 (1995).
  17. Footnote 42.Corley v. United States, 556 U.S. 303, 314 (2009) (“[A] statute should be construed so that effect is given to all its provisions, so that no part will be inoperative or superfluous, void or insignificant.”) (internal quotation marks and citation omitted).
  18. Footnote 51.People-First Chatbot Act § 2(a)(1)(C)(i).
  19. Footnote 60.Jiang Lu et al., Large Language Models and Their Applications in Drug Discovery and Development: A Primer, 18 Clinical & Translational Sci. no. 4 (2025), https://doi.org/10.1111/cts.70205 (last visited July 15, 2026)
  20. Footnote 60.Mauro Giuffrè et al., Systematic Review: The Use of Large Language Models as Medical Chatbots in Digestive Diseases, 60 Alimentary Pharmacology & Therapeutics 144 (2024), https://doi.org/10.1111/apt.18058 (last visited July 15, 2026).
  21. Footnote 61.Ole Pütz, Co-Textual Dopes: How LLMs Produce Contextually Appropriate Text in Chat Interactions with Humans Without Access to Context, 36 J. Linguistic Anthropology no. 1 (2025), https://doi.org/10.1111/jola.70036 (last visited July 15, 2026).
  22. Footnote 62.Pengfei He et al., Towards the Effect of Examples on In-Context Learning: A Theoretical Case Study, 14 Stat no. 1 (2025), https://doi.org/10.1002/sta4.70045 (last visited July 15, 2026).
  23. Footnote 64.People-First Chatbot Act § 4(c)(1).
  24. Footnote 71.16 C.F.R. § 312.2 (definition of personal information, as amended)
  25. Footnote 71.Children’s Online Privacy Protection Rule, 90 Fed. Reg. 16,977 (Apr. 22, 2025) (final amendments; full compliance required by April 22, 2026).
  26. Footnote 72.16 C.F.R. § 312.5(a)(2).
  27. Footnote 73.16 C.F.R. § 312.5(a)(1).
  28. Footnote 74.16 C.F.R. § 312.5(c)(9).
  29. Footnote 76.People-First Chatbot Act § 4(c)(1)(B)(i).
  30. Footnote 82.Wilson Contreras-Higuera, Lucrezia Crescenzi-Lanna & Mirko Duradoni, The Role of Time in Facial Dynamics and Challenges in Automatic Emotion Recognition (2019–2024), 2025 Hum. Behav. & Emerging Techs. no. 1, tbl. 8, https://doi.org/10.1155/hbe2/7777949 (last visited July 15, 2026).
  31. Footnote 85.Manju Priya Ramaswamy & Suja Palaniswamy, Multimodal Emotion Recognition: A Comprehensive Review, Trends, and Challenges, 14 WIREs Data Mining & Knowledge Discovery no. 6 (2024), https://doi.org/10.1002/widm.1563 (last visited July 15, 2026).
  32. Footnote 86.People-First Chatbot Act § 5(19)(A).
  33. Footnote 87.Goldman v. Healthcare Mgmt. Sys., Inc., 559 F. Supp. 2d 853, 868 (W.D. Mich. 2008) (“Section 1203(c)(3) authorizes statutory damages ‘for each violation’ of sections 1201 and 1202, without defining what constitutes a violation.”).
  34. Footnote 91.740 Ill. Comp. Stat. 14/20.
  35. Footnote 92.Cothron v. White Castle Sys., Inc., 2023 IL 128004, ¶ 1.
  36. Footnote 94.Pub. Act 103-0769 (eff. Aug. 2, 2024)
  37. Footnote 94.see, e.g., McGivney v. ITS Techs. & Logistics, LLC, 2025 IL App (1st) 241961-U, ¶ 22.
  38. Footnote 96.United States v. Epic Games, Inc., No. 5:22-cv-00518 (E.D.N.C. Feb. 7, 2023) (stipulated order for permanent injunction and civil penalty)
  39. Footnote 96.United States v. Edmodo, LLC, No. 3:23-cv-02495 (N.D. Cal. June 27, 2023) (same).
  40. Footnote 97.Bert-Jaap Koops, The Concept of Function Creep, 13 L. Innovation & Tech. 29, 29 (2021), https://doi.org/10.1080/17579961.2021.1898299 (last visited July 15, 2026).
  41. Footnote 98.Sarah B. Evans-Jordan & John-Arne Skolbekken, Scientific Citizenship’s Youngest Domain: Function Creep in Norway’s Newborn Screening Programme, 26 Sci. Tech. & Soc’y 98, 101 (2021) (“The mere fact of storage provides a logistical opportunity, if not a regulatory one, for unrelated secondary use.”), https://doi.org/10.1177/0971721820964891 (last visited July 15, 2026).
  42. Footnote 100.Wakefield v. ViSalus, Inc., 51 F.4th 1109, 1118, 1122 (9th Cir. 2022).
  43. Footnote 101.TransUnion LLC v. Ramirez, 594 U.S. 413, 426–27 (2021)
  44. Footnote 101.Spokeo, Inc. v. Robins, 578 U.S. 330, 341 (2016) (same principle, remanding for concreteness analysis).
  45. Footnote 102.Gadelhak v. AT&T Servs., Inc., 950 F.3d 458, 462–63 (7th Cir. 2020) (Barrett, J.).
  46. Footnote 104.15 U.S.C. § 6502(a)(1).
  47. Footnote 106.Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 Apr. 2016, arts. 4(1), 8, 2016 O.J. (L 119) 1, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32016R0679 (last visited July 15, 2026).
  48. Footnote 107.15 U.S.C. § 6502(c)–(d).
  49. Footnote 108.Adjustments to Civil Penalty Amounts, 90 Fed. Reg. 5,580 (Jan. 17, 2025)
  50. Footnote 108.Off. of Mgmt. & Budget, Exec. Off. of the President, M-26-11 (Apr. 17, 2026).
  51. Footnote 108.No Adjustment to Civil Monetary Penalty Amounts, 91 Fed. Reg. 41,722 (July 7, 2026).
  52. Footnote 109.740 Ill. Comp. Stat. 14/10, 14/20
  53. Footnote 109.Cothron, 2023 IL 128004, ¶ 38 (the statutory violation itself is the injury, and a plaintiff “is not required to plead or prove actual damages”) (citing Rosenbach v. Six Flags Ent. Corp., 2019 IL 123186).
  54. Footnote 110.Tex. Bus. & Com. Code § 503.001(d) (enforcement by the attorney general)
  55. Footnote 110.Wash. Rev. Code § 19.375.030 (enforcement through the consumer protection act).
  56. Footnote 111.Press Release, Office of Rep. Valerie Foushee, Reps. Foushee, Casar Introduce Legislation to Protect Children and Americans’ Privacy from AI Chatbot Harms (July 9, 2026), https://foushee.house.gov/media/press-releases/reps-foushee-casar-introduce-legislation-to-protect-children-and-americans-privacy-from-ai-chatbot-harms-and-require-chatbot-safety-assessments (endorser statement describing the bill as one that “protects state laws and the families already fighting these companies”).
  57. Footnote 112.18 U.S.C. §§ 2510(2), 2511(2)(d).
  58. Footnote 113.Cal. Penal Code § 632(a), (c).
  59. Footnote 114.740 Ill. Comp. Stat. 14/10, 14/15(b).
  60. Footnote 118.People-First Chatbot Act § 2(b)(2)(A).
  61. Footnote 121.People-First Chatbot Act § 3(b).
  62. Footnote 122.Loper Bright Enters. v. Raimondo, 603 U.S. 369 (2024).

All research