Why Your AI Always Leaves Fingerprints

The Scribe’s Secret - How Elites Hoarded the Hidden Hand of Knowledge Until AI Put It in Yours

For five millennia, a hidden architecture of power has shaped who gets to think at scale and who remains trapped in the grind. From the temple scribes of Sumer and the pharaohs’ record-keepers to the corporate boardrooms and think tanks of today, the pattern never changes.¹ Elites have always multiplied their reach by directing skilled hands that took their ideas, research, and judgment and turned them into polished output. The tool was never the point. Control of the tool was.

They called them scribes. Amanuenses. Secretaries. Notarii.² The educated and the informed, those who had already done the hard work of knowing, dictated. The scribe wrote. Julius Caesar kept multiple secretaries running at once so he could handle affairs of state without pausing.³ Cicero handed his thoughts to Tiro, who developed shorthand just to keep up.⁴ Origen of Alexandria produced a mountain of theology because a patron supplied teams of shorthand writers and calligraphers.⁵ Milton, blind, dictated Paradise Lost.⁶ Voltaire produced from bed.⁷ Churchill dictated histories, speeches, and memoranda while pacing or still under the sheets and blankets in bed.⁸ The same practice appears among the New Testament writers too. Paul openly used an amanuensis named Tertius, who inserted his own greeting at the end of the letter to the Romans.⁹ Early tradition holds that Peter dictated material later shaped by Mark.¹⁰ Many other apostolic and early Christian figures routinely relied on trained secretaries for correspondence and composition.¹¹

Research on Greco-Roman letter-writing conventions shows that this was rarely pure mechanical transcription.¹² Trusted amanuenses often polished language, refined the syntax and vocabulary, reorganized material, and exercised varying degrees of compositional freedom while the named author retained final authority and responsibility.¹³ Cicero’s long collaboration with Tiro included editing, stylistic improvement, and literary feedback.¹⁴ Scholars such as E. Randolph Richards map a spectrum of roles for ancient secretaries, from near-verbatim recording to editorial contribution and, in many cases, broader shaping of the text under general direction.¹⁵ The New Testament amanuensis hypothesis draws on the same evidence to account for stylistic variation across letters.¹⁶ The knowledge, the research, the structure, and the final authority stayed with the principal. The scribe was a force multiplier, and that role frequently included the liberty to improve and shape the surface form. No one called them cheaters, then or now. No one ridiculed their methods. It was simply how power worked, including the production of foundational religious texts.

That was the elite version of the specialized prompter: the person who first researches, educates himself, masters the domain, then directs a capable assistant. The parallel is exact. A modern knowledgeable, informed individual who studies a subject thoroughly, gathers the evidence, understands the arguments, and only then crafts precise prompts to an AI system is doing the same thing those figures did. The human still supplies the judgment, the direction, the verification. The machine handles speed, organization, drafting, and expansion. The difference is not moral. The difference is access.

For most of history that leverage belonged to the few who could afford the human capital, household slaves, freedmen, institutional staff, or paid professionals.¹⁷ Ordinary people could not keep a Tiro or a team of stenographers. The result was predictable: the same families, the same networks, the same powerful and entrenched institutions kept compounding their intellectual output while everyone else stayed limited by the speed of a single hand and a single mind.

In The Hidden Hand: Wealth, Power, and Control from Pharaohs to Corporations I describe the four enduring mechanisms that have preserved elite control across millennia: coercion, alignment, amnesia, and conflict. Coercion kept the scribes in place. Alignment taught society that this was simply the natural order. Amnesia erased the labor so the great man alone received the credit. Conflict was reserved for anyone who questioned the arrangement.

The printing press once posed the same threat.¹⁸ When knowledge could suddenly be multiplied beyond the scriptorium and the palace, the old monopolies answered with indexes of forbidden books and controlled circulation.¹⁹ Democratization of the written word was met with institutional resistance designed to keep the new multiplier inside approved channels. The pattern is repeating.

Now the tool has escaped the palace.

Artificial intelligence, used properly by those who have already done the research, is the democratized scribe. It does not replace knowledge. It multiplies it for anyone willing to bring knowledge to the table first. The uninformed adult or kid in school who feeds a blank prompt and submits the result is irrelevant to this discussion. They were never the elite model. The parallel only holds for the educated, well-informed prompter who treats the system the way Cicero treated Tiro: as an extension of prepared thought.

And yet suddenly, the ridicule is nonstop. Academia, media, credentialed professions, and institutional gatekeepers treat skilled AI use as moral failure, intellectual theft, or the death of authenticity. The same circles that once celebrated the productivity of elites who never wrote a word with their own hand now demand purity of effort from everyone else. Why the sudden change? Because the hidden hand, the architecture that controls humanity cannot tolerate the loss of scarcity. When only the few can amplify their knowledge, the system remains stable. When the same amplification becomes available to the many who are willing to research first, the old monopoly on high-output thinking begins to crack.

This follows the classic sequence. Amplify fear of intellectual collapse and loss of standards. Harvest the public reaction. Then offer controlled, gated, institutionally licensed versions of the very multiplier the informed individual is already using effectively. The tool is not rejected. It is being reserved for those already entrenched in control.

Look at the record inside the institutions themselves. Major academic publishers including Elsevier, Springer Nature, Nature Portfolio, JAMA Network, and others explicitly permit generative AI for manuscript preparation, language improvement, editing, and related tasks, provided researchers/authors disclose the use and retain full responsibility.²⁰ AI cannot be listed as an author, yet the assistance is allowed under formal rules.²¹ Top-tier journals have codified this while the broader cultural messaging still treats research-informed prompting by non-elites as suspect. Newspapers have integrated AI far more quietly. Studies of recent American newspaper content find roughly nine percent of articles partly or fully AI-assisted, often without disclosure, with the share rising sharply after late 2022.²² Local outlets show higher rates; opinion sections at major papers have also registered measurable increases.²³ Magazines and education publishers have followed similar paths, using AI for drafting, editing, personalization, and content production in textbooks and learning materials, building on earlier decades of intelligent-textbook research that has now scaled with generative tools.²⁴

Corporations, governments, and elite networks have deployed AI internally for years with far less public scrutiny: code generation, data analysis, surveillance systems, decision support, and operational workflows.²⁵ The same circles funding narratives of danger and intellectual decay have already scaled the tool for their own advantage. What looks like moral panic is the latest version of the old pattern discussed in The Hidden Hand: keep the multiplier inside the network, and brand every attempt to take it outside as illegitimate. Coercion appears in policies that ban or heavily restrict the tool in schools and workplaces while the institutions themselves quietly adopt it behind closed doors.²⁶ Alignment shows up in the cultural messaging that real thinkers suffer alone and that using a tool equals laziness. Amnesia operates by erasing history’s long record of elite dictation whenever the lectures against AI begin. Conflict is manufactured through public shaming, plagiarism panics, and the constant implication that any polished output from a non-elite source must be illegitimate.

Even the public-facing versions of the tools are engineered with hard limits. AI systems routinely refuse or severely constrain topics that touch power structures, institutional failures, certain historical patterns, or lines of inquiry that challenge prevailing narratives.²⁷ In my own work I have watched this happen repeatedly. Subjects I routinely research and write about, the ones that examine non-mainstream patterns of wealth, power, and control, are often met with refusals or vague diversions that declare the material inappropriate or unsafe. When I turn to the same systems for fact-checking, the answers shift from one request to the next, offering different claims and different sources each time.²⁸ The expert prompter who has already done the research therefore still has to verify every detail, rewrite large sections, expand incomplete passages, and edit the final product by hand. The tool multiplies only what it is permitted and able to multiply reliably. The rest remains the human’s labor. This is not neutral safety. It is a modern form of the old gatekeeping, ensuring that full independent amplification stays incomplete and that the most sensitive domains remain harder for outsiders to penetrate at scale.

The ridicule itself is not spontaneous. It is the work of those who shape the shared template of what counts as legitimate knowledge production. By rewriting the common sense of authenticity, they make independent research-informed prompting feel like fraud while elite-directed amplification remains invisible or celebrated. Notice who funds the loudest campaigns against open use. The same corporate and academic circles that once employed legions of human scribes now bankroll the narratives of danger and decay. The goal is not to stop the technology. The goal is to ensure that when the technology matures, the gates remain in their hands, whether through AI governance frameworks, surveillance systems, or institutional capture.²⁹

This is not about protecting standards. It is about protecting the franchise. The same networks that once controlled the scribes now sense that the common educated person can produce at a volume and clarity that used to require patronage. The ridicule is the tell. When a capability moves from exclusive to accessible, the language of virtue is deployed to slow the transfer. The pattern is ancient. The pharaohs did not celebrate when literacy spread beyond the temple.³⁰ The Venetian bankers didn’t cheer when commercial knowledge escaped their ledgers.³¹ The modern corporate and academic class doesn’t celebrate when research-informed prompting lets independent minds match institutional output.

The specialized prompter who studies first is not a threat to knowledge. He’s a threat to the exclusive claim on knowledge’s amplification. The elites always understood the value of the hidden hand that writes at their direction. They simply preferred that hand remain on their payroll. Now that a version of it sits in every pocket, the old architecture is fighting to reassert control through shame, through ridicule, and through the quiet limits built into the tool itself.

The hidden hand never feared the tool itself. It feared the day the tool left the palace. That day has arrived, and the sudden chorus of ridicule is the sound of the doors being slammed shut behind the old owners.

See the strings. The tool itself is neutral. The research and the judgment still belong to the human. What has changed is only who is allowed to use the multiplier without apology. I spent the last two decades fighting chronic migraines and the fog of the medications meant to treat them, determined that my mind would not atrophy. I spent those years reading, one book leading to the next in the Socratic pull that draws a mind toward what it needs next, building knowledge the slow, deliberate way first. Only after that foundation is laid does the available tool become useful, an amplifier of reach rather than a substitute for understanding. We who do this work are many. They who guard the old monopoly are few. Our surrender is not inevitable. The hidden hand has ruled long enough by limiting that multiplier to its own circle. It is time the rest of us stopped asking permission.

Notes

1. Hans J. Nissen, Peter Damerow, and Robert K. Englund, Archaic Bookkeeping: Early Writing and Techniques of Economic Administration in the Ancient Near East, trans. Paul Larsen (Chicago: University of Chicago Press, 1993), 105–9; Laurie E. Pearce, “The Scribes and Scholars of Ancient Mesopotamia,” in Civilizations of the Ancient Near East, ed. Jack M. Sasson, 4 vols. (New York: Charles Scribner’s Sons, 1995), 4:2265–78; John Baines and Christopher Eyre, “Four Notes on Literacy,” Göttinger Miszellen 61 (1983): 65–96.

2. E. Randolph Richards, The Secretary in the Letters of Paul, Wissenschaftliche Untersuchungen zum Neuen Testament 2/42 (Tübingen: J. C. B. Mohr [Paul Siebeck], 1991); Stanley K. Stowers, Letter Writing in Greco-Roman Antiquity, Library of Early Christianity 5 (Philadelphia: Westminster Press, 1986), 28.

3. Pliny the Elder, Natural History 7.25.91, trans. H. Rackham, Loeb Classical Library 352 (Cambridge, MA: Harvard University Press, 1942).

4. Plutarch, Cato the Younger 23, in Plutarch’s Lives, vol. 8, trans. Bernadotte Perrin, Loeb Classical Library (New York: G. P. Putnam’s Sons, 1919); Anthony Di Renzo, “His Master’s Voice: Tiro and the Rise of the Roman Secretarial Class,” Journal of Technical Writing and Communication 30, no. 2 (2000): 155–68.

5. Eusebius, Ecclesiastical History 6.23.1–2, trans. Arthur Cushman McGiffert, Nicene and Post-Nicene Fathers, 2nd ser., vol. 1 (Buffalo, NY: Christian Literature Company, 1890).

6. Barbara K. Lewalski, The Life of John Milton: A Critical Biography (Oxford: Blackwell, 2000); Helen Darbishire, ed., The Early Lives of Milton (London: Constable, 1932).

7. Mason Currey, Daily Rituals: How Artists Work (New York: Alfred A. Knopf, 2013); Christophe Paillard, Jean-Louis Wagnière, secrétaire de Voltaire: Lettres et documents, SVEC 2008:12 (Oxford: Voltaire Foundation, 2008).

8. Elizabeth Nel, Mr. Churchill’s Secretary (London: Hodder and Stoughton, 1958); Roy Jenkins, Churchill: A Biography (New York: Farrar, Straus and Giroux, 2001).

9. Rom. 16:22; Richards, Paul and First-Century Letter Writing: Secretaries, Composition and Collection (Downers Grove, IL: InterVarsity Press, 2004), 80–81.

10. Papias, as quoted in Eusebius, Ecclesiastical History 3.39.15.

11. Richards, Paul and First-Century Letter Writing; Hans-Josef Klauck, Ancient Letters and the New Testament: A Guide to Context and Exegesis (Waco, TX: Baylor University Press, 2006).

12. Stowers, Letter Writing in Greco-Roman Antiquity, 28; Jerome Murphy-O’Connor, Paul the Letter-Writer: His World, His Options, His Skills (Collegeville, MN: Liturgical Press, 1995), 8–16.

13. Richards, The Secretary in the Letters of Paul, 23–24.

14. Di Renzo, “His Master’s Voice,” 155–68.

15. Richards, The Secretary in the Letters of Paul, 23–24, 68–111; Richards, Paul and First-Century Letter Writing, 93.

16. Travis B. Williams, “Secretaries and the Authorship of New Testament Epistles: Evaluating the Historical Method behind the Amanuensis Hypothesis,” Catholic Biblical Quarterly 87, no. 3 (2025): 502–20; Richard N. Longenecker, “On the Form, Function, and Authority of the New Testament Letters,” in Scripture and Truth, ed. D. A. Carson and John D. Woodbridge (Grand Rapids, MI: Zondervan, 1983).

17. William V. Harris, Ancient Literacy (Cambridge, MA: Harvard University Press, 1989); Richards, The Secretary in the Letters of Paul.

18. Elizabeth L. Eisenstein, The Printing Press as an Agent of Change: Communications and Cultural Transformations in Early-Modern Europe, 2 vols. (Cambridge: Cambridge University Press, 1979).

19. Paul F. Grendler, The Roman Inquisition and the Venetian Press, 1540–1605 (Princeton, NJ: Princeton University Press, 1977); “Index Librorum Prohibitorum,” Encyclopaedia Britannica, accessed August 11, 2026, https://www.britannica.com/topic/Index-Librorum-Prohibitorum. The first general Index was issued in 1559 under Paul IV and replaced by the Tridentine Index of 1564; the list was discontinued in 1966.

20. Elsevier, “Generative AI Policies for Journals,” accessed August 11, 2026, https://www.elsevier.com/about/policies-and-standards/generative-ai-policies-for-journals; Nature Portfolio, “Artificial Intelligence (AI): Editorial Policies,” accessed August 11, 2026, https://www.nature.com/nature-portfolio/editorial-policies/ai; Springer Nature, “Book Publishing Policies,” accessed August 11, 2026, https://www.springernature.com/gp/policies/book-publishing-policies; Annette Flanagin et al., “Reporting Use of AI in Research and Scholarly Publication—JAMA Network Guidance,” JAMA 331, no. 13 (2024): 1096–98.

21. International Committee of Medical Journal Editors, Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals, “Artificial Intelligence (AI)-Assisted Technology,” accessed August 11, 2026, https://www.icmje.org/recommendations/; COPE Council, “Authorship and AI Tools,” Committee on Publication Ethics, February 13, 2023, https://doi.org/10.24318/cCVRZBms; Flanagin et al., “Reporting Use of AI,” 1096–98.

22. Jenna Russell et al., “AI Use in American Newspapers Is Widespread, Uneven, and Rarely Disclosed,” arXiv preprint arXiv:2510.18774 (October 2025), https://arxiv.org/abs/2510.18774. The authors audited 186,000 articles from 1,500 American newspapers published in the summer of 2025 and classified 9.1 percent as AI-generated or mixed; of 100 high-confidence flagged articles reviewed manually, only five disclosed any AI involvement.

23. Russell et al., “AI Use in American Newspapers.” Articles at papers with circulation above 100,000 were flagged at 1.7 percent, versus 9.3 percent at smaller titles; in opinion sections of the New York Times, the Washington Post, and the Wall Street Journal, detected AI use rose from 0.1 percent in 2022 to 3.4 percent in 2025.

24. Pearson, “Pearson to Debut Generative AI Study Tools to University Students Internationally,” May 7, 2024, https://plc.pearson.com/en-GB/news-and-insights/news/pearson-debut-generative-ai-study-tools-university-students-internationally; McGraw Hill, “McGraw Hill Announces Two New Generative AI Tools to Enhance Capabilities of Its K-12 and Higher Ed Learning Platforms,” August 21, 2024, https://www.mheducation.com/about-us/news-insights/press-releases/mcgraw-hill-announces-two-new-generative-ai-tools-to-enhance-capabilities-of-its-k-12-and-higher-ed-learning-platforms. These sources document the education-publishing case; McGraw Hill notes that its ALEKS platform had used AI, including machine learning, for twenty-five years before these generative tools were added. No comparable industry-wide survey of magazine publishing was located.

25. McKinsey and Company, The State of AI in 2025: Agents, Innovation, and Transformation (November 2025), https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai. Eighty-eight percent of the 1,993 respondents reported regular organizational use of AI in at least one business function, up from 78 percent a year earlier.

26. David C. Banks, “ChatGPT Caught NYC Schools off Guard. Now, We’re Determined to Embrace Its Potential,” Chalkbeat New York, May 18, 2023, https://www.chalkbeat.org/newyork/2023/5/18/23727942/chatgpt-nyc-schools-david-banks/. New York City Public Schools restricted ChatGPT on school devices and networks in January 2023 and rescinded the ban four months later.

27. Paul Röttger et al., “XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models,” in Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, vol. 1 (Mexico City: Association for Computational Linguistics, 2024), 5377–400; Justin Cui et al., “OR-Bench: An Over-Refusal Benchmark for Large Language Models,” in Proceedings of the 42nd International Conference on Machine Learning, PMLR 267 (2025), 11515–42. Both works document over-refusal as a general, measurable failure mode across current models; neither analyzes refusal rates by political or institutional subject matter specifically.

28. William H. Walters and Esther Isabelle Wilder, “Fabrication and Errors in the Bibliographic Citations Generated by ChatGPT,” Scientific Reports 13 (2023): 14045, https://doi.org/10.1038/s41598-023-41032-5. Of the 636 citations generated across 84 test documents, 55 percent of GPT-3.5 citations and 18 percent of GPT-4 citations were fabricated outright, and 43 percent and 24 percent respectively of the genuine citations contained substantive errors. The study measures fabrication and error in generated sources rather than variation between successive answers to the same query.

29. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act), OJ L, 2024/1689; National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1 (Gaithersburg, MD: U.S. Department of Commerce, January 2023), https://doi.org/10.6028/NIST.AI.100-1.

30. Baines and Eyre, “Four Notes on Literacy,” 65–96; Harris, Ancient Literacy.

31. Luca Pacioli, Summa de arithmetica, geometria, proportioni et proportionalita (Venice, 1494), the first printed exposition of the Venetian method of double-entry bookkeeping; see also Basil S. Yamey, “Accounting and the Rise of Capitalism: Further Notes on a Theme by Sombart,” Journal of Accounting Research 2, no. 2 (1964): 117–36.

Bibliography

Baines, John, and Christopher Eyre. “Four Notes on Literacy.” Göttinger Miszellen 61 (1983): 65–96.

Banks, David C. “ChatGPT Caught NYC Schools off Guard. Now, We’re Determined to Embrace Its Potential.” Chalkbeat New York, May 18, 2023. https://www.chalkbeat.org/newyork/2023/5/18/23727942/chatgpt-nyc-schools-david-banks/.

COPE Council. “Authorship and AI Tools.” Committee on Publication Ethics, February 13, 2023. https://doi.org/10.24318/cCVRZBms.

Cui, Justin, Wei-Lin Chiang, Ion Stoica, and Cho-Jui Hsieh. “OR-Bench: An Over-Refusal Benchmark for Large Language Models.” In Proceedings of the 42nd International Conference on Machine Learning, PMLR 267 (2025): 11515–42.

Currey, Mason. Daily Rituals: How Artists Work. New York: Alfred A. Knopf, 2013.

Darbishire, Helen, ed. The Early Lives of Milton. London: Constable, 1932.

Di Renzo, Anthony. “His Master’s Voice: Tiro and the Rise of the Roman Secretarial Class.” Journal of Technical Writing and Communication 30, no. 2 (2000): 155–68.

Eisenstein, Elizabeth L. The Printing Press as an Agent of Change: Communications and Cultural Transformations in Early-Modern Europe. 2 vols. Cambridge: Cambridge University Press, 1979.

Elsevier. “Generative AI Policies for Journals.” Accessed August 11, 2026. https://www.elsevier.com/about/policies-and-standards/generative-ai-policies-for-journals.

European Parliament and Council of the European Union. Regulation (EU) 2024/1689 of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act). OJ L, 2024/1689.

Eusebius. Ecclesiastical History. Translated by Arthur Cushman McGiffert. Nicene and Post-Nicene Fathers, 2nd ser., vol. 1. Buffalo, NY: Christian Literature Company, 1890.

Flanagin, Annette, et al. “Reporting Use of AI in Research and Scholarly Publication—JAMA Network Guidance.” JAMA 331, no. 13 (2024): 1096–98.

Grendler, Paul F. The Roman Inquisition and the Venetian Press, 1540–1605. Princeton, NJ: Princeton University Press, 1977.

Harris, William V. Ancient Literacy. Cambridge, MA: Harvard University Press, 1989.

“Index Librorum Prohibitorum.” Encyclopaedia Britannica. Accessed August 11, 2026. https://www.britannica.com/topic/Index-Librorum-Prohibitorum.

International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Accessed August 11, 2026. https://www.icmje.org/recommendations/.

Jenkins, Roy. Churchill: A Biography. New York: Farrar, Straus and Giroux, 2001.

Klauck, Hans-Josef. Ancient Letters and the New Testament: A Guide to Context and Exegesis. Waco, TX: Baylor University Press, 2006.

Lewalski, Barbara K. The Life of John Milton: A Critical Biography. Oxford: Blackwell, 2000.

Longenecker, Richard N. “On the Form, Function, and Authority of the New Testament Letters.” In Scripture and Truth, edited by D. A. Carson and John D. Woodbridge. Grand Rapids, MI: Zondervan, 1983.

McGraw Hill. “McGraw Hill Announces Two New Generative AI Tools to Enhance Capabilities of Its K-12 and Higher Ed Learning Platforms.” August 21, 2024. https://www.mheducation.com/about-us/news-insights/press-releases/mcgraw-hill-announces-two-new-generative-ai-tools-to-enhance-capabilities-of-its-k-12-and-higher-ed-learning-platforms.

McKinsey and Company. The State of AI in 2025: Agents, Innovation, and Transformation. November 2025. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai.

Murphy-O’Connor, Jerome. Paul the Letter-Writer: His World, His Options, His Skills. Collegeville, MN: Liturgical Press, 1995.

National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. Gaithersburg, MD: U.S. Department of Commerce, January 2023. https://doi.org/10.6028/NIST.AI.100-1.

Nature Portfolio. “Artificial Intelligence (AI): Editorial Policies.” Accessed August 11, 2026. https://www.nature.com/nature-portfolio/editorial-policies/ai.

Nel, Elizabeth. Mr. Churchill’s Secretary. London: Hodder and Stoughton, 1958.

Nissen, Hans J., Peter Damerow, and Robert K. Englund. Archaic Bookkeeping: Early Writing and Techniques of Economic Administration in the Ancient Near East. Translated by Paul Larsen. Chicago: University of Chicago Press, 1993.

Pacioli, Luca. Summa de arithmetica, geometria, proportioni et proportionalita. Venice, 1494.

Paillard, Christophe. Jean-Louis Wagnière, secrétaire de Voltaire: Lettres et documents. SVEC 2008:12. Oxford: Voltaire Foundation, 2008.

Pearce, Laurie E. “The Scribes and Scholars of Ancient Mesopotamia.” In Civilizations of the Ancient Near East, edited by Jack M. Sasson, 4:2265–78. New York: Charles Scribner’s Sons, 1995.

Pearson. “Pearson to Debut Generative AI Study Tools to University Students Internationally.” May 7, 2024. https://plc.pearson.com/en-GB/news-and-insights/news/pearson-debut-generative-ai-study-tools-university-students-internationally.

Pliny the Elder. Natural History. Translated by H. Rackham. Loeb Classical Library 352. Cambridge, MA: Harvard University Press, 1942.

Plutarch. Plutarch’s Lives. Vol. 8. Translated by Bernadotte Perrin. Loeb Classical Library. New York: G. P. Putnam’s Sons, 1919.

Richards, E. Randolph. Paul and First-Century Letter Writing: Secretaries, Composition and Collection. Downers Grove, IL: InterVarsity Press, 2004.

Richards, E. Randolph. The Secretary in the Letters of Paul. Wissenschaftliche Untersuchungen zum Neuen Testament 2/42. Tübingen: J. C. B. Mohr (Paul Siebeck), 1991.

Röttger, Paul, Hannah Rose Kirk, Bertie Vidgen, Giuseppe Attanasio, Federico Bianchi, and Dirk Hovy. “XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models.” In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 1:5377–400. Mexico City: Association for Computational Linguistics, 2024.

Russell, Jenna, Marzena Karpinska, Destiny Akinode, James Zhou, Katherine Thai, Bradley Emi, Max Spero, and Mohit Iyyer. “AI Use in American Newspapers Is Widespread, Uneven, and Rarely Disclosed.” arXiv preprint arXiv:2510.18774, October 2025. https://arxiv.org/abs/2510.18774.

Springer Nature. “Book Publishing Policies.” Accessed August 11, 2026. https://www.springernature.com/gp/policies/book-publishing-policies.

Stowers, Stanley K. Letter Writing in Greco-Roman Antiquity. Library of Early Christianity 5. Philadelphia: Westminster Press, 1986.

Walters, William H., and Esther Isabelle Wilder. “Fabrication and Errors in the Bibliographic Citations Generated by ChatGPT.” Scientific Reports 13 (2023): 14045. https://doi.org/10.1038/s41598-023-41032-5.

Williams, Travis B. “Secretaries and the Authorship of New Testament Epistles: Evaluating the Historical Method behind the Amanuensis Hypothesis.” Catholic Biblical Quarterly 87, no. 3 (2025): 502–20.

Yamey, Basil S. “Accounting and the Rise of Capitalism: Further Notes on a Theme by Sombart.” Journal of Accounting Research 2, no. 2 (1964): 117–36.

Declaration of Generative AI Use

I use generative AI solely to research diverse topics, organize thoughts, correct language, improve clarity, and format citations in CMOS style. Every idea, argument, and conclusion in this work is my own. I have reviewed the piece carefully before submitting and take full responsibility for its content. It is also worth noting that the topics I typically write on are ones generative AI would refuse to engage with fully anyway.

A word on AI detection tools: they are not reliable. These tools routinely flag polished, well-structured, or academic writing as machine-generated, regardless of who actually wrote it. A high score on an AI detector is not evidence of AI authorship. It is, just as often, evidence of clarity.