The Deleted Face: Why Facial Recognition Must Erase You First

August 3, 2026

On a January afternoon in 2020, Robert Williams pulled into the driveway of his home in Farmington Hills, Michigan, and was arrested on his own front lawn while his wife and two young daughters watched. Detroit police accused him of stealing watches from a Shinola boutique. The case against him was, in essence, a single thing: a grainy frame of in-store surveillance footage that an algorithm had decided looked like the photograph on his driver's licence. He was held for roughly thirty hours in an overcrowded cell, made to sleep on a concrete floor, and questioned over a crime committed by a man he had never met and did not resemble in any way that a human eye, given a moment of honest attention, would have confirmed. When detectives finally laid the surveillance still beside his face, even one of them seemed to concede the obvious. The computer, Williams later recalled being told, must have got it wrong.

It is a story that has, by now, hardened into a parable. Williams was the first person in the United States known to have been wrongfully arrested because of a face recognition match. He would not be the last. Porcha Woodruff, eight months pregnant, was arrested in Detroit in February 2023 for a carjacking and held for around eleven hours, though nothing in the surveillance or witness accounts described a visibly pregnant woman; the photo lineup put before the victim used an eight-year-old mugshot rather than her current driver's licence photograph. Her charges were dismissed. Nijeer Parks spent ten days in a New Jersey jail for a shoplifting and assault he could not have committed, having been thirty miles away making a money transfer at the time. Robert Dillon, a fifty-two-year-old from Fort Myers, Florida, was arrested in August 2024 for allegedly trying to lure a child from a fast-food restaurant in Jacksonville Beach, a city he had never visited, three hundred miles from home, after police ran a grainy image of the suspect through an AI-assisted facial recognition system that returned him at 93 per cent “confidence”. Charged with a third-degree felony, he saw the case dropped more than two months later, once his attorney showed he had been at work. The ACLU and the ACLU of Florida sued on his behalf on 10 June 2026. By the ACLU's tally there are now at least fifteen such cases. Nearly all of the wrongfully arrested were Black.

The familiar way to tell this story is as a tale of error. The system, we say, made a mistake. It misidentified. The accuracy was poor, the dataset unrepresentative, the threshold miscalibrated. Fix the maths, broaden the training data, audit the vendors, and the harm recedes. This is the framing of most policy debate, most journalism, and a good deal of the technical literature. It is also, argues a paper presented at the 2026 ACM Conference on Fairness, Accountability, and Transparency and published in its proceedings, a profound misreading of what these systems actually do.

The paper, which appears under the title “Frankenstein in the Pipeline: Computational Epistemicide in Facial Recognition” and circulates as the arXiv preprint 2606.07628, makes a claim more unsettling than miscalibration. Its author, the Brazilian computer scientist Nina da Hora, takes Mary Shelley's creature not as a parable of unintended consequences but as a description of method: a body disassembled, reassembled from parts, legitimated by the procedure that made it. Facial recognition, da Hora contends, does not merely misidentify people from Black and non-Western communities. It performs something closer to an act of erasure. Through a sequence of ordinary engineering steps, the technology takes the face as a living, relational surface and progressively narrows it to whatever can be held still as data, then measures the residue against a norm that is, in its statistical bones, predominantly white, frontal, and European. To be recognised by such a system, the argument runs, anyone whose face departs from that norm must first be remade in its image. Da Hora gives this process a deliberately heavy name: computational epistemicide. The killing, by computation, of a way of being known.

It is a phrase designed to make you flinch, and it should. But before deciding whether it is overheated, it is worth doing something the policy conversation rarely does. It is worth looking, carefully and without squeamishness, at what actually happens to a face when a machine sets out to recognise it.

The Pipeline That Eats a Face

A modern face recognition system is not a single model that gazes at you and knows your name. It is an assembly line, and like all assembly lines it works by subtraction. At each station, something is removed, normalised, or thrown away, until what remains is a thing that can travel.

The first station is detection. Before a system can recognise a face it must find one, and finding means deciding where, in a rectangle of pixels, a face begins and ends. A detector returns a bounding box, a confidence score, and usually a handful of coarse keypoints. Already a decision has been made about what counts. A face partially turned away, shadowed, veiled, dark against a dark background, or simply lit in a way the detector's training did not anticipate may not register as a face at all. The earliest and most quietly consequential form of exclusion is not being misidentified. It is being invisible to the camera in the first place, falling below the threshold at which the machine agrees that a person is present.

The second station is cropping. The detected region is excised from its surroundings. The context goes: the body, the setting, the people standing alongside, the weather of the moment. What had been a person in a world becomes a rectangle of skin and feature. This is the first amputation, and it is so banal that no engineer would think to call it one. Yet a face is not, in lived experience, a free-floating object. It is always a face turned towards or away from someone, in a place, doing something. The crop dissolves all of that as a precondition of proceeding.

The third station is landmarking. The system locates fiducial points: the corners of the eyes, the tip and base of the nose, the edges of the mouth, the line of the jaw. These coordinates are the skeleton on which everything downstream depends. Landmark detectors are trained on annotated faces, and the geometry they expect, the assumptions about where features should sit relative to one another, carries the statistical signature of the data they learned from. A face whose proportions, expression, or pose sit outside that learned distribution yields noisier, less confident landmarks, and the error propagates.

The fourth station, and the one where the paper's argument bites hardest, is alignment. Having found the landmarks, the system warps the face. It applies a geometric transformation, rotation, scaling, translation, sometimes more aggressive distortion, so that the eyes sit on a predefined horizontal line, the nose falls on a fixed axis, the whole face is dragged into a canonical frontal pose at a standard size. The destination of that warp is a template, a set of target coordinates representing where a face is supposed to be. Every face entering the system is bent towards the same template. And that template was not handed down from nature. It was derived, historically and statistically, from the faces that dominated the field's foundational datasets, which were overwhelmingly white, male, and photographed front-on. Alignment is the moment, in the most literal mechanical sense, when your face is reshaped to fit a norm that did not come from you.

The fifth and final station is embedding. The aligned crop is fed into a deep convolutional network, which collapses it into a vector: a list of numbers of fixed length. The landmark FaceNet system, published by Florian Schroff, Dmitry Kalenichenko, and James Philbin at Google in 2015, mapped each face to a compact point in a Euclidean space, originally just 128 bytes per face, where distance between points stood in for similarity between faces. Later systems commonly use 512 dimensions. The output is L2-normalised, scaled to unit length, so that what survives of you is a direction in a high-dimensional space, a fixed-dimensional artefact stripped of scale, context, history, and flesh.

That vector is the point of the whole exercise. It is small, portable, and comparable. It can be stored in a database, indexed, matched against a watchlist, shipped between agencies, and queried in milliseconds. It is, in the engineering sense, beautiful: an entire human face rendered as a few hundred numbers that you can do arithmetic on. The cosine of the angle between two such vectors becomes a verdict on whether two faces belong to the same person.

Read the pipeline back as a single motion and the paper's central image comes into focus. A face enters as a living surface and exits as a coordinate. At each step the criterion is the same, mostly unspoken: keep what can be stabilised as data, discard the rest. Vectorisation, in da Hora's reading, completes the stitching: the dissected face sewn back together as a fixed-dimensional artefact whose purpose is to circulate. The face is not photographed so much as it is metabolised.

What the Face Was Before the Machine

To feel the force of the word epistemicide, you have to take seriously what the pipeline is subtracting, and that means refusing, for a moment, the engineer's flat definition of a face as a region of an image.

A face is not, primarily, a static object to be measured. It is a surface in motion and in relation. It is the principal instrument through which human beings recognise, address, and answer one another. Philosophers have long argued that the face is the very site of ethical demand, the place where another person confronts you as a person and not a thing. Across cultures the face is bound up with honour, shame, kinship, deference, and belonging. We speak of saving face and losing face, of facing someone, of a face that falls. None of this is metaphor laid over a neutral biological substrate. It is what the face actually is in the lives of the people who wear one. The face is relational before it is anatomical.

The recognition pipeline cannot hold any of that, and it is important to be precise about why. The problem is not that engineers are careless. It is that the relational face is, by definition, the part that will not stand still. It changes with whom you are addressing and how you feel about them. It is constituted in the encounter. A system whose entire purpose is to produce a stable, transmissible token must, of structural necessity, treat everything relational as noise to be normalised away. Alignment exists precisely to cancel pose, expression, and angle, which is to say to cancel the face as an act and preserve only the face as a fingerprint.

This is where the borrowed word does real work. Epistemicide was coined in 1995 by the Portuguese sociologist Boaventura de Sousa Santos, who used it to name the destruction of entire systems of knowing, the indigenous, southern, and subaltern ways of understanding the world that colonial modernity did not merely defeat but rendered illegitimate, unthinkable, gone. Santos's wager in his book Epistemologies of the South is that there can be no global social justice without what he calls cognitive justice, a recognition of the many valid ways human beings come to know. Epistemicide is what happens when one way of knowing installs itself as the only way, and the others are not argued with but erased.

But the lineage the paper actually claims runs through a second thinker, and it is the more pointed one. In a 2005 doctoral thesis at the University of São Paulo, the Black Brazilian philosopher Sueli Carneiro took Santos's term and turned it on her own country, using it to name the way Black Brazilians are stripped of standing as legitimate subjects of knowledge: disqualified in advance as knowers, constructed as the other who is not, so that the question of what they know need never arise. Da Hora's computational epistemicide is explicitly an extension of Carneiro's. Which is why the racial argument does not have to be bolted on afterwards: it is what the philosophy was built to describe.

Da Hora's manoeuvre is to apply that lineage to the face. The claim is that facial recognition enacts a small, mechanised epistemicide every time it runs. There is a way of knowing a person that is relational, embodied, reciprocal, the way a face is known by those who love it or live beside it. And there is the way of knowing installed by the pipeline, in which a person is a vector calibrated against a canonical norm. The second does not coexist with the first. In the systems that decide who boards, who enters, and who is flagged, it replaces it. The relational face is not weighed and found wanting. It is simply not represented in the data artefact at all. What gets killed is not the person. It is a way of the person being known.

The Norm Has a Demographic

You could grant all of this as philosophy and still object that it floats free of the engineering. Surely, the objection runs, a vector is just a vector. The numbers do not know your race. Here the empirical record is unkind to the objection, because the canonical norm the paper describes is not an abstraction. It has been measured, and it leaves fingerprints in the error rates.

In 2018, Joy Buolamwini and Timnit Gebru published Gender Shades, an audit of commercial gender-classification systems from IBM, Microsoft, and the Chinese company Face++. The results were stark. For lighter-skinned men, error rates sat under one per cent. For darker-skinned women, they rose as high as 34.7 per cent, with some disaggregated figures worse still. The systems worked best on the faces that most resembled the people and datasets they had been built around, and degraded precisely as faces departed from that centre. It was not a uniform fog of inaccuracy. It was a gradient, and the gradient had a colour.

A year later the US National Institute of Standards and Technology, the federal body that runs the authoritative Face Recognition Vendor Test, published its demographic study, evaluating scores of algorithms from across the industry. Its findings have become the empirical backbone of the entire debate. In one-to-one matching, the kind used to verify that you are who your passport says you are, the systems produced false positives for Asian and African American faces at rates ranging, depending on the algorithm, from ten to one hundred times higher than for white faces. In one-to-many matching, the kind a police force uses to search a face against a database of mugshots, African American women were among those carrying elevated false-positive rates, as NIST's own summary noted. The picture is more tangled than a headline permits: on the mugshot imagery the highest rates fell on Native American faces, with African American and Asian faces also elevated, and the ordering shifts by sex and by algorithm. The disparity is unmistakable; its precise shape is not uniform. NIST also noted a revealing wrinkle: some algorithms developed in Asian countries did not show the same penalty against Asian faces, strong evidence that the disparity tracks the composition of the data and the norm baked into it, not anything intrinsic to the faces themselves.

That last point matters enormously for the paper's thesis. If the bias were a fixed property of how cameras meet melanin, you would expect every system everywhere to fail in the same direction. It does not. The norm is contingent. It is built, and it could in principle be built differently, which is exactly why calling it a norm rather than a law is correct. The canonical face, frontal and pale, is an artefact of which faces happened to fill the foundational datasets and define the alignment templates of a field that grew up in particular institutions, in particular countries, photographing particular people. Everyone else is rendered legible only by being warped towards a centre that was never theirs, and the cost of that warping shows up, with grim reliability, as a higher chance of being confused with a stranger.

This is the bridge between the philosophical claim and the statistical one. The disparate error rate is not a separate problem from the epistemicide. It is the epistemicide becoming visible. The false match is the moment the system's insistence on remaking you in the image of its norm fails loudly enough to land you in a cell.

The Stakes Are No Longer Hypothetical

For most of facial recognition's history, the argument over what it does could remain somewhat academic, because the technology sat mostly at the edges of consequential life. That window has closed. The face is now a credential at the most heavily guarded thresholds in modern society, and the decisions that turn on it are precisely the ones from which there is least room to appeal.

Begin with the border. In December 2025, a US Department of Homeland Security final rule took effect authorising Customs and Border Protection to collect facial biometrics from all non-citizens on both entry and exit, by air, land, and sea. Earlier exemptions for children under fourteen and adults over seventy-nine were stripped away, so that virtually every non-citizen crossing a US frontier is now photographed and matched. The Transportation Security Administration has been pushing facial verification through the domestic system too, though more haltingly than the coverage implies: as of 2026 its facial-matching programme runs at roughly sixty-five airports, with face-scanning units at around eighty-four and more than two hundred and fifty lanes accepting digital identification. The figure of four hundred-plus airports that circulates in reporting on the rollout is a target for late 2026, not a fact, and a TSA spokesperson has conceded that full operation may not arrive until 2030, or even 2040. CBP, for its part, expects full deployment of biometric exit across commercial airports and seaports within three to five years. The face is becoming the boarding pass, the passport, and the turnstile, and the population most exposed to it, by the explicit design of the rule, is non-citizens, the very group whose faces the NIST data shows the systems handle worst.

Then there is policing, where the Williams, Woodruff, Parks, and Dillon cases are not anomalies but the visible tip of a practice. A face is captured on surveillance, run against a database of millions, and returned as a ranked list of candidates. Too often that lead, which the vendors themselves caution is merely investigative, is treated as probable cause, and a name at the top of a list becomes a knock at the door. The harm is distributed exactly where the error rates predict it will be. The systems fail most on Black faces, and it is Black people who keep being arrested for crimes committed by someone the machine decided they resembled.

The institutional appetite is widening from there: building access, benefits fraud screening, examination invigilation, age verification, retail loss prevention. In each case the seductive promise is the same frictionless certainty, your face as a key that cannot be forgotten or lent out. And in each case the same translation occurs beneath the surface. You are detected, cropped, landmarked, aligned, and embedded, and the entity that is actually admitted or refused is not you but your vector, measured against a norm.

Regulators have begun, unevenly, to respond, and the shape of their response reveals how narrowly the problem is still being understood. The European Union's AI Act, whose first prohibitions took effect in February 2025, bans real-time remote biometric identification in public spaces for law enforcement, subject to carved-out exceptions for finding missing persons, preventing imminent threats to life or terrorist attacks, and locating suspects in serious crimes. It also prohibits building face databases by untargeted scraping of the internet or CCTV, the practice that made the company Clearview AI notorious. These are real and meaningful limits. But notice their grammar. They regulate where and when and against whom the pipeline may be pointed. They do not touch what the pipeline does to a face once it is pointed. Crucially, retrospective identification, analysing footage after the event, is treated merely as high risk rather than banned.

And the high-risk regime that would have governed it has just slipped. The EU's Digital Omnibus on AI entered into force on 27 July 2026, postponing the obligations attaching to the Annex III high-risk categories, which expressly cover biometrics, law enforcement, and border management, from August 2026 to 2 December 2027; Annex I obligations move to 2 August 2028. The outright prohibitions were not touched: what Article 5 forbade in February 2025 it forbids still. But the machinery that was to have disciplined retrospective facial identification, risk management, technical documentation, data governance, human oversight, conformity assessment, has been deferred by well over a year, while the cameras and the contracts go in regardless. The law governs deployment, and even that has been rescheduled. The translation from person to vector proceeds untouched.

What Is Actually Lost in Translation

So return to the question the paper forces, the one the accuracy debate keeps stepping around. When a face is reduced to a fixed-dimensional artefact built to circulate across databases and institutions, the issue is not only whether the system works. It is what is lost in the translation, and who pays for the loss.

Three things are lost, and they compound.

The first is context, and with it the very possibility of relation. The relational face, the face as address and answer, is precisely the part the pipeline must discard to do its job. A face turned in greeting, a face set in defiance, a face slack with grief: these are erased at alignment, because alignment exists to cancel exactly such variation. What the system preserves is the part of you that holds still, which is the least human part, the part most like a barcode. To be recognised by the machine is to be recognised only as the thing in you that does not change, and never as the person doing the changing.

The second is consent over your own legibility, and this is more radical than the familiar complaint about privacy. The standard privacy worry is that the system sees too much of you. The deeper worry the paper surfaces is that to be seen at all you must first be rewritten. Anyone whose face departs from the canonical norm is made legible only at the price of being warped towards it, remade in the image of a centre that was never theirs. You do not get to be recognised as yourself. You get to be recognised as a deviation from someone else, measured by how far you had to be bent to fit. That is a strange and corrosive form of recognition, one that withholds the very thing the word promises.

The third loss is recourse. A vector circulates. Once your face has been embedded and entered into a database, the artefact travels between agencies, jurisdictions, and private vendors at machine speed, decoupled from the moment and the body it was taken from. If it is matched in error, the error propagates with the same efficiency. You cannot easily see the vector, cannot inspect it, cannot correct it, and in most jurisdictions cannot compel its deletion. The thing standing in for you in the rooms where decisions are made is one you have never been shown and cannot answer. When it speaks against you, as it spoke against Robert Williams, the burden of disproof falls on the living person, who must somehow argue with a number.

And the courts have offered only a patchy remedy. Williams reached a landmark settlement with Detroit in 2024 that obliged the department to rewrite how it may act on face recognition leads. Porcha Woodruff got the opposite: in August 2025 a federal judge, calling her arrest and jailing “troubling for many reasons”, nonetheless dismissed her civil rights claim against the officer who prepared her warrant, holding that her lawyers had not shown he lacked probable cause. Her attorney said he was shocked by the decision and intended to appeal. The same technology, the same city, the same pattern of harm, and two entirely different answers to the question of whether anyone is answerable. If you cannot see the artefact that accused you, and cannot reliably sue those who acted on it, recourse has been hollowed out while remaining on the books.

As for who bears the cost: the empirical record settles it without ambiguity. The translation is not equally lossy for everyone. It costs most where faces depart furthest from the canonical norm, which is to say it costs Black people, non-Western people, women with darker skin, the very populations in whose name Carneiro reworked the concept of epistemicide. The technology recapitulates, in silicon and at scale, the older pattern in which one way of knowing installs itself as universal and bills everyone else for the privilege of being misread by it.

Being Recognised by Something That Cannot See You

Robert Williams has said that what stayed with him was not only the wrongful night in a cell but the strange affront of it, the sense of having been confused with a man he was not, by a process he could not interrogate, in front of the children to whom he is most fully and irreplaceably himself. His daughters know his face in the way a face is actually known, as the living surface of the person who comes home. The system that arrested him knew a vector, aligned to a template, and decided that vector was close enough to another to be worth a warrant.

The gap between those two ways of knowing is the whole of the matter. One is relational, reciprocal, and irreducible to a coordinate. The other is portable, comparable, and built for circulation through the institutions that increasingly stand between people and the places they need to go. Da Hora's wager is that we have spent a decade arguing about whether the second kind of knowing is accurate, when the prior and harder question is whether it should be permitted to stand in for the first at all, in the rooms where it now does.

To be recognised by a system that can only see you by first making you someone else is, the paper suggests, not really to be recognised at all. It is to be replaced, at the threshold, by a more convenient version of yourself: stilled, flattened, calibrated against a norm you did not set and may never have matched. For a growing share of the decisions that govern a life, who boards, who enters, who is flagged, who is freed, that replacement is becoming the default condition of being seen. The deleted face is not a malfunction at the edge of the technology. It is the technology working exactly as designed. The unsettling achievement of the paper is to make us ask, while the rollout is unfinished and the four hundredth airport still an ambition, whether a design that must delete the face in order to read it is one we should be installing at the doors of public life at all.

References

American Civil Liberties Union, “Williams v. City of Detroit: Face Recognition False Arrest.” https://www.aclu.org/cases/williams-v-city-of-detroit-face-recognition-false-arrest
Michigan Public Radio, “'It didn't make sense at all': Wrongful facial recognition arrest in Detroit leads to landmark settlement,” 28 June 2024. https://www.michiganpublic.org/criminal-justice-legal-system/2024-06-28/it-didnt-make-sense-at-all-wrongful-facial-recognition-arrest-leads-to-landmark-settlement
American Civil Liberties Union, “More than a Dozen Wrongful Arrests Due to Police Reliance on Facial Recognition Technology.” https://www.aclu.org/news/privacy-technology/more-than-a-dozen-wrongful-arrests-due-to-police-reliance-on-facial-recognition-technology
NBC News, “Detroit woman sues city after being falsely arrested while pregnant due to facial recognition technology.” https://www.nbcnews.com/news/us-news/detroit-woman-sues-city-falsely-arrested-8-months-pregnant-due-facial-rcna98447
Joy Buolamwini and Timnit Gebru, “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification,” Proceedings of Machine Learning Research, vol. 81, 2018. https://proceedings.mlr.press/v81/buolamwini18a.html
MIT News, “Study finds gender and skin-type bias in commercial artificial-intelligence systems,” 12 February 2018. https://news.mit.edu/2018/study-finds-gender-skin-type-bias-artificial-intelligence-systems-0212
National Institute of Standards and Technology, “NIST Study Evaluates Effects of Race, Age, Sex on Face Recognition Software,” 19 December 2019. https://www.nist.gov/news-events/news/2019/12/nist-study-evaluates-effects-race-age-sex-face-recognition-software
National Institute of Standards and Technology, Face Recognition Vendor Test (FRVT) Part 3: Demographic Effects (NISTIR 8280). https://pages.nist.gov/frvt/reports/demographics/
Florian Schroff, Dmitry Kalenichenko and James Philbin, “FaceNet: A Unified Embedding for Face Recognition and Clustering,” CVPR 2015, arXiv:1503.03832. https://arxiv.org/abs/1503.03832
Boaventura de Sousa Santos, Epistemologies of the South: Justice Against Epistemicide (Routledge, 2014). https://www.routledge.com/Epistemologies-of-the-South-Justice-Against-Epistemicide/Santos/p/book/9781612055459
U.S. Customs and Border Protection, “DHS announces Final Rule to advance the Biometric Entry/Exit Program.” https://www.cbp.gov/newsroom/national-media-release/dhs-announces-final-rule-advance-biometric-entry/exit-program
Federal Register, “Collection of Biometric Data From Aliens Upon Entry to and Departure From the United States,” 27 October 2025. https://www.federalregister.gov/documents/2025/10/27/2025-19655/collection-of-biometric-data-from-aliens-upon-entry-to-and-departure-from-the-united-states
Biometric Update, “TSA targets 400 US airports for biometrics rollout,” December 2024. https://www.biometricupdate.com/202412/tsa-targets-400-us-airports-for-biometrics-rollout
European Parliament, “EU AI Act: first regulation on artificial intelligence.” https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence
Future of Privacy Forum, “Red Lines under the EU AI Act: Restricting Real-time Remote Biometric Identification Systems for Law Enforcement Purposes.” https://fpf.org/blog/red-lines-under-the-eu-ai-act-restricting-real-time-remote-biometric-identification-systems-for-law-enforcement-purposes/
Nina da Hora, “Frankenstein in the Pipeline: Computational Epistemicide in Facial Recognition,” Proceedings of the 2026 ACM Conference on Fairness, Accountability, and Transparency (FAccT '26), Montreal, June 2026. https://doi.org/10.1145/3805689.3812284
Nina da Hora, “Frankenstein in the Pipeline: Computational Epistemicide in Facial Recognition,” arXiv:2606.07628. https://arxiv.org/abs/2606.07628
American Civil Liberties Union, “Dillon v. City of Jacksonville Beach.” https://www.aclu.org/cases/dillon-v-city-of-jacksonville-beach
American Civil Liberties Union, “Florida Man Sues Police Over Wrongful Arrest Due to False Facial Recognition Match,” 10 June 2026. https://www.aclu.org/press-releases/florida-man-sues-police-over-wrongful-arrest-due-to-false-facial-recognition-match
Sueli Carneiro, A Construção do Outro como Não-Ser como fundamento do Ser (doctoral thesis, University of São Paulo, 2005).
Gibson Dunn, “EU AI Act Omnibus Agreement — Postponed High-Risk Deadlines and Other Key Changes.” https://www.gibsondunn.com/eu-ai-act-omnibus-agreement-postponed-high-risk-deadlines-and-other-key-changes/
CBS News Detroit, “Woman wrongly accused of carjacking loses lawsuit against Detroit police who used facial technology,” 4 September 2025. https://www.cbsnews.com/detroit/news/woman-wrongly-accused-carjacking-loses-lawsuit-detroit-police-used-facial-tech/

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Listen to the free weekly SmarterArticles Podcast

Discuss...