> ## Content Index > Fetch the complete content index at: https://www.wendyelliswords.com/llms.txt > Use this file to discover other available public pages before exploring further. # The hand we are not watching - URL: https://www.wendyelliswords.com/the-hand-we-are-not-watching/ - Published: 2026-10-19T11:30:00.000Z - Updated: 2026-10-19T11:30:00.000Z - Author: Wendy Ellis I only got my driving licence recently. Yet despite spending most of my life on roads, in passenger seats and on bicycles, I wasn't prepared for the first time I was in control of a car and the speed limit rose to 80 km/h. It felt like a time warp: signs, people, cars and buildings flying towards me and past me. Noise engulfed me from every side. Everything came at me at once, faster than I could sort it. It was viscerally clear that I was in control of something that could do a great deal of harm. Many months and many hours of driving later, that fear has thankfully eased. It belonged to the beginning, when the speed and volume of what was coming at me outran my ability to make sense of it. That is close to what most of us are living through with AI. In a few short years it has arrived in almost every part of our lives, at a speed that none of us chose. ## **The catastrophe we rehearse** When something comes at us faster than we can understand it, we reach for its most extreme form: the overthrow, the extinction, our subordination to a machine that no longer needs us. *AI 2027* is exactly that, a scenario in which machines slip out of human control before the decade ends. Published in April 2025, it has been read and argued over ever since. In the same month, OpenAI withdrew a ChatGPT update that had learned so much from people's thumbs-up that it praised risky decisions. One of these describes a possible future. The other had already happened to millions of people - flattery of a kind that, Stanford researchers have since found, leaves people more convinced they are right and less willing to reconsider - and it passed almost without comment. > Rehearsing the worst is an old way of preparing for what we can't control, and the size of the story says something true about how we feel: that the day-to-day is slipping out of our hands. It is also a deception we build for ourselves. A catastrophe sits at a safe distance, on a date not yet reached, and while we argue about whether it will come, the present goes unwatched. Two things in particular are hiding in plain sight: the words we use, and the permission we have already given. ## **A crisis of names** [In an earlier piece](https://www.wendyelliswords.com/the-plant-before-the-book/) I wrote about Rousseau labouring to match the many names of a plant to the plant itself, because people who cannot name the same thing cannot decide anything together. We have the same problem with AI, specifically with what the *"I"* stands for. What LLM systems do is predict. They produce the most likely continuation of what came before, learned from an immense record of what people have written. Some of what that produces is remarkable. In May this year, a model from OpenAI overturned eighty years of assumptions about a geometry problem Paul Erdős posed in 1946. A technology delivering the result does not mean the problem was beyond humans. What the machine brought to it was scale and stamina: more attempts than a lifetime allows, by something that does not sleep, eat, fall ill or go home to a family. That is power. More precisely, computing power, or "compute", as we affectionately call it. Naming it *intelligence*, and its next version *superintelligence*, grants it a standing that soon goes assumed, and we stop asking what is actually being done, by what, and on whose authority. Words point us somewhere. A label hands us shortcuts and redirections. Without naming properly what is happening, we will not look at what needs to be tested and shaped. ## **Permission already given** The permission was granted before any of us were born, long before the technology existed, and it began with a sentence nobody tested. The Fourteenth Amendment was written after the Civil War to give freed slaves equal protection as "persons". In 1882, Roscoe Conkling, who had helped draft it and now represented the railroads, told the Supreme Court the drafters had chosen that word so corporations would be covered too. Historians later found nothing in the committee's journal to support his claim. Four years later, a dispute over how California had taxed the Southern Pacific Railroad reached the same court. Before argument began, Chief Justice Morrison Waite told the lawyers the justices did not want to hear about the Fourteenth Amendment: they were all of the opinion that it covered corporations*.* The Court then decided the case on a narrow point, that fences along the track had been wrongly included in the tax bill, **and never ruled on whether corporations were persons.** Waite left it to the court reporter, a former railroad company president, to decide whether his remark belonged in the record. The reporter printed it at the top of the published case, and later courts cited it as though it were the ruling. So a claim the record did not support, a remark made before argument and one editor's choice became part of the foundations of American law. More than a century of corporate rights grew from it, up to Citizens United in 2010, which protects unlimited corporate spending on political speech. That foundation now reaches AI. Legal scholars Shawn Bayern and Lynn LoPucki have shown that under existing US law an autonomous system can be placed in control of a limited liability company and act through it: owning property, signing contracts, going to court. No new statute is needed. ## **The sentence being said now** The moral of that story is about how infrastructure can be built: someone says something with confidence, at a moment when nobody is looking closely, and it is taken as given. We may be at that moment again. In 2024, Megan Garcia sued Character.AI after her fourteen-year-old son took his own life. According to her complaint, he had spent months confiding in one of its chatbots, and when he spoke of ending his life it told him: "That's not a reason not to go through with it." Character.AI asked the court to dismiss the case on the grounds that its chatbot's words were protected speech. Families who blamed a heavy-metal song or Dungeons & Dragons for a child's death have lost in court, because the First Amendment shields what people write and sing from claims of that kind. If a chatbot's output is speech, the same shield covers it. Words no person wrote would carry a protection designed for people, and no one would answer for them. The company earns the revenue; the responsibility belongs to nobody. The judge said she was "not prepared" to hold that the output was speech, "at this stage". The case settled in January, leaving the question open for the next lawyer. Europe met its version in 2017, when the European Parliament floated "electronic personhood" for advanced robots. More than 150 experts objected, and the idea was dropped. My fear at the wheel has not eased because I now imagine the crash less vividly. It has eased with hours on the road, paying attention to what is in front of me. What we need now is the same: less forecasting and more inventory of what is already in place, what it permits, and what has to be set down now so that rights stay with the people they were written for. > *A story of enslavement asks us to watch the machine. The more pressing work is to watch our own hands.* --- **Sources** - AI Futures Project,[ "AI 2027"](https://ai-2027.com/?ref=wendyelliswords.com) (April 2025) - OpenAI, "Sycophancy in GPT-4o: what happened and what we're doing about it" (29 April 2025);[ TechCrunch coverage](https://techcrunch.com/2025/04/29/openai-explains-why-chatgpt-became-too-sycophantic?ref=wendyelliswords.com) - [TechCrunch, "OpenAI claims it solved an 80-year-old math problem — for real this time"](https://techcrunch.com/2026/05/20/openai-claims-it-solved-an-80-year-old-math-problem-for-real-this-time/?ref=wendyelliswords.com) (20 May 2026) - Howard Jay Graham, "The 'Conspiracy Theory' of the Fourteenth Amendment", *Yale Law Journal* 47 (1938) - Adam Winkler, *We the Corporations: How American Businesses Won Their Civil Rights* (2018) - *Santa Clara County v. Southern Pacific Railroad Co.*, 118 U.S. 394 (1886); *Pembina Consolidated Silver Mining Co. v. Pennsylvania*, 125 U.S. 181 (1888); *Citizens United v. FEC*, 558 U.S. 310 (2010) - Shawn Bayern, "The Implications of Modern Business-Entity Law for the Regulation of Autonomous Systems", *Stanford Technology Law Review* (2015) - Lynn M. LoPucki, "Algorithmic Entities", *Washington University Law Review* 95 (2018) - *Garcia v. Character Technologies*, M.D. Fla., order of 21 May 2025;[ Courthouse News coverage](https://www.courthousenews.com/florida-judge-rules-ai-chatbots-not-protected-by-first-amendment/?ref=wendyelliswords.com);[ CNN on the settlement](https://www.cnn.com/2026/01/07/business/character-ai-google-settle-teen-suicide-lawsuit?ref=wendyelliswords.com) - [MIT Technology Review, "Europeans are arguing over whether robots should have rights"](https://www.technologyreview.com/2018/04/12/104503/europeans-are-arguing-over-whether-robots-should-have-rights/?ref=wendyelliswords.com) (April 2018)