Dolly Parton and the Tribute Flood: How Fakes Rewrite Memory

On Friday 4 September 2026, ten days after her sister died, Stella Parton opened Instagram and wrote a message to several million strangers. Parts of it were warm. She thanked people for what she called genuine support during the family's bereavement. Then came the part that travelled, picked up by NBC News, TODAY and HuffPost among others across the following seventy-two hours. “There is a tremendous amount of AI and endless garbage being posted everyday and it's been challenging to absorb and or ignore,” she wrote. She criticised those “in the business of exploiting the loss and tragedies of others,” who would, she predicted, simply “move on to the next stampede of endless misinformation.” She asked people to show more compassion. She said the family had experienced incredible pressure and insensitivity from strangers.

Dolly Parton died on 25 August 2026 in Nashville, aged eighty. Her publicist, Marcel Pariseau, gave the cause as a brief battle with cancer. The family statement described a rhinestone life that shone bright enough for the world to see. That is the ordinary architecture of a public bereavement, unchanged since the wire services learned to move faster than the funeral.

What had not existed before was the thing Stella Parton was describing. Within forty-eight hours, the internet had begun manufacturing Dolly Parton. Not remembering her. Manufacturing her. On the morning of 27 August, Mashable found at least three YouTube videos carrying fabricated thumbnails of Parton in a hospital gown, tubes taped to her face, sitting in the platform's Videos on the Rise and YouTube Featured shelves, some amplified through the Hype button YouTube built to help small creators find an audience. Three of the four carried a Made with AI label reading: sounds or visuals were altered or fully generated. It sat under a picture of a woman dying in a bed she never lay in.

That same day, the President of the United States posted an AI-generated illustration to Truth Social showing Dolly Parton on his arm, walking a White House colonnade. No photograph of the two of them together appears to exist. The image was not a memory. It was a wish rendered at photographic fidelity, and it drew criticism from fans who read it as an attempt to annex a woman famous for refusing political endorsement. Songs followed. The BBC identified one, a track presented as a heartfelt fan-made tribute from Parton's goddaughter Miley Cyrus, as an AI generation rather than a performance. It had passed 1.5 million views.

Nobody has counted the total, and it is worth saying so rather than reaching for a number that does not exist. There is no audit of how many synthetic Dolly Parton images, songs or videos appeared between 25 August and 4 September 2026. There is a Mashable reporter who found three videos on one morning, a BBC identification of one song with seven figures of views, a presidential post, a widely shared fake photograph purporting to show Parton without makeup or wig, and a sister who called the aggregate endless. That is the verified record, and its thinness is part of the problem: producing this material is trivially cheap and measuring it is not.

Ten Days of Machine-Made Condolence

What makes the Parton case legible is that almost every participant would describe themselves as a mourner. The channels putting a dying woman on a thumbnail are grief farms, but the comment sections beneath them are full of people typing rest in peace. The AI Miley Cyrus song was received by many listeners as devotion from a goddaughter to a godmother. Some of the people who shared it were moved. They cried at a vocal performance that no larynx produced, about a woman neither the singer nor the machine had met.

This is the gap the Parton family fell into. Between the tribute the sender believes they are offering and the thing the recipient receives there is now a distance wide enough to lose a person in. The sender experiences a moment of feeling, a prompt, a post. Maybe fifteen seconds. The recipient experiences a feed.

It is tempting to blame bad actors, and the bad actors are there. But grief farming and grief are not separable populations here. They use the same tools, occupy the same shelves in the same recommendation systems, and produce artefacts that look identical. A sincere fan's generated portrait of Parton ascending into clouds and a content farm's are, to the sister scrolling past them, the same object. The intent is invisible. Only the output arrives.

The Rehearsal Nobody Recognised

Eleven months before she died, Dolly Parton had to deny her own death.

In early October 2025, an AI-generated image spread showing Parton in a hospital bed with a breathing tube, Reba McEntire seated beside her, praying and wiping away tears. It was fabricated. Parton, then seventy-nine, had postponed a run of Las Vegas shows for medical procedures, and the internet had filled the silence. On 8 October she posted a video to Instagram to say she was not dying. She was funny about it. Both of them, she said, looked like they needed to be buried, and if she were really dying she did not think Reba would be the one at her deathbed.

Read from the far side of August 2026, the episode is less funny. The machinery that produced the deathbed image while she was alive produced the hospital thumbnails after she died. The only variable that changed was whether anyone was left to laugh it off. A living subject can post a video and say, I ain't dead yet. A dead one cannot. What a family inherits along with the estate is the job of denial, and denial does not scale the way generation does.

There is a further wrinkle. Parton had already, on the record, declined the digital afterlife. Asked at a London press conference in July 2023 whether she would consider a hologram show of the kind ABBA and the Whitney Houston estate had staged, she said she had left behind a great body of work and had to decide how much of that high-tech stuff she wanted to be involved with, “because I don't want to leave my soul here on this earth.” It was an offhand remark, not a legal instrument, and also about as close to an expressed premortem preference as any performer of her generation has given. It was contradicted within forty-eight hours of her death by people who believed they were honouring her.

Three Harms Wearing the Same Coat

The public argument about synthetic tributes collapses three separate injuries into one indignant blur. Keeping them apart is the only way to work out which parts of this are new.

The first harm is offence to the bereaved. Stella Parton's complaint is, on its face, about being subjected to something unwanted during the worst fortnight of her life. This is real, and it is not unprecedented. Families of the famous dead have always absorbed intrusion, from the graveside photographer to the tribute act booking a venue before the funeral. What is new is not the category but the volume. You can decline an interview. You cannot decline a feed.

The second harm is corruption of the public record. An image of Donald Trump arm in arm with Dolly Parton at the White House depicts an encounter for which no photographic evidence exists. It is now indexed, screenshotted and reposted, and it will be retrievable long after the context of its creation has evaporated. Every synthetic artefact enters the same searchable substrate as the real archive, and the systems that rank them were not built to distinguish an event from a rendering of one. This has precedents, from Stalin's airbrushed commissars to the doctored magazine cover, but those were scarce and required institutional access. Now they cost nothing.

The third harm is the alteration of private memory, and it has no precedent at scale. The claim is that exposure to synthetic depictions of a person can change what the people who knew them actually recall. It sounds like the sort of assertion that gets made in a feature about AI and never substantiated. In this instance it has been tested.

What Happened in a Room Where the Photographs Had Been Edited

In September 2024 a team from the MIT Media Lab published a pre-registered experiment titled “Synthetic Human Memories: AI-Edited Images and Videos Can Implant False Memories and Distort Recollection,” later published in the proceedings of the 2025 CHI conference. The authors were Pat Pataranutaporn, Chayapatr Archiwaranguprok, Samantha W. T. Chan, Elizabeth Loftus and Pattie Maes. Loftus is the reason the paper carries weight: she spent five decades establishing that memory is reconstructive rather than archival, that misleading post-event information reshapes what witnesses believe they saw, and that fabricated childhood episodes can be installed in a substantial minority of ordinary adults. In 2007, with Dario Sacchi and Franca Agnoli, she showed that doctored photographs of real public events changed how Italian participants recalled them; those shown an altered image of a Rome anti-war march rated it as more violent and recalled more confrontation, property damage and injury than participants who saw the original. The 2024 study asks what happens when the doctoring is automatic, free and universally available.

The design was simple. Two hundred US participants, aged twenty to seventy-three, viewed twenty-four unedited photographs for two minutes, then played Pac-Man as a distractor. Depending on which of four groups they had been assigned to, they then saw the same unedited images, a set in which twelve had been altered using Adobe's generative editing tools, five-second videos animated from the unedited stills using Luma's Dream Machine, or videos animated from the AI-edited stills. Afterwards they answered questions about the originals and rated their confidence.

In the control group, 18.88 per cent of answers were false memories. After AI-edited stills, that rose to 31.37 per cent. After AI-generated video built from AI-edited stills, it reached 38.67 per cent, a little over twice the control rate, and correct recollections fell from 66.16 per cent to 43.83 per cent. The differences were statistically robust.

The confidence data is the part that should worry anyone. Participants in the video-of-edited-image condition rated their false memories at 5.41 out of seven, against 4.54 in the control group. They were more certain whilst being wrong. Nor was the effect defeated by disclosure: the authors note that false memories still formed significantly where labels indicated an image had been enhanced, which they take as evidence that passive notification is insufficient as a cognitive safeguard. Younger participants were somewhat more susceptible. Scepticism towards media, familiarity with AI filters and self-rated memory ability made no significant difference at all.

Two details matter for the Parton case. The edits that produced the highest absolute rate of false memory were edits to people, where the strongest condition induced false recollections in 45.3 per cent of responses. And the authors raise the application themselves, discussing the TikTok trend of animating photographs of dead relatives and writing that it “may blur the line between genuine memories and digitally fabricated ones, potentially affecting how people remember their loved ones.”

Honesty requires marking the limits. This was a laboratory study of strangers looking at unfamiliar photographs for two minutes, not a study of siblings recalling six decades of shared life, and memories of someone you actually loved are encoded far more deeply than a Pac-Man-interrupted glance at a stock image. Extrapolating from one to the other is an inference rather than a finding. The direction of that inference is not seriously in doubt, though. In March 2026 Yifan Li and Xingyu Lan published interviews with twenty-six bereaved users of chatbots built to simulate dead relatives, and found a progression: the bot initially reinforces memory, then the generated interactions begin to blur with authentic recollection. The authors flag memory distortion as an explicit ethical concern.

Why the Mural Comparison Falls Apart

The strongest objection to everything above is that none of it is new, and it deserves a proper hearing rather than a strawman.

Amateur tribute has always attended celebrity death. Murals appear on gable ends within days. Fans paint portraits, write songs, stage vigils. Impersonators have made careers out of the dead since the first Elvis competition, and religious art has for two millennia depicted people in scenes the artists did not witness without anyone arguing that an altarpiece corrupts the record. The dead have always been imagined. Imagining them is arguably the central function of mourning, which is why continuing bonds theory, set out by Dennis Klass, Phyllis Silverman and Steven Nickman in 1996, reframed healthy grief around maintaining a relationship with the deceased rather than severing one.

Four things have changed, and they compound.

Volume. The marginal cost of a tribute has fallen to roughly zero. A mural takes a week and a ladder. A generated portrait takes four seconds. Any system in which supply is unbounded and attention is not will saturate, and saturation is what Stella Parton meant by endless.

Speed. Synthetic production outruns verification by an order of magnitude. The hospital thumbnails were live on 27 August. The BBC identification of the fake Cyrus song came later, after 1.5 million views. Fact-checking is a serial human process; generation is a parallel machine one. The asymmetry is structural and will not close.

Fidelity. This is the crucial one. A mural announces itself as a mural. Its style is a signature of human hands, and a viewer decodes it as homage. A photorealistic generation borrows the epistemic authority of the camera, an authority built over a century and a half on the assumption that light had to bounce off something for the image to exist. That assumption is now false, and no equivalent cultural competence has replaced it.

Searchability. The mural stays on its wall. The generation enters the index, gets scraped, gets ranked against genuine archive material, and may end up as training data for the next model. Fan art occupied a separate space from the record. Synthetic tribute occupies the same space and is retrieved by the same queries.

Sincerity Is a Fact About the Sender

Defenders of synthetic tribute make an argument worth taking seriously: these people mean it. Denying them an expressive medium because a family finds it distasteful is a kind of gatekeeping, and grief policing has its own unpleasant literature. Whoever generated an image of Dolly Parton in heaven with her late husband Carl Dean was not committing an act of aggression. They were doing what mourners do, which is to picture the dead somewhere better.

The defence fails, but not for the reason usually given.

It fails because sincerity is a claim about the sender's interior state, whilst tribute is a claim about what the recipient receives, and the two come apart routinely. The sender is not lying about their feelings. The feelings simply do not travel with the artefact, and the artefact is the only thing that arrives.

There is also a specific mechanism by which generative tools break tribute, and it concerns cost. In every gift economy anthropologists have studied, the effort embedded in a gift is part of its meaning. The hand-drawn portrait communicates devotion because it consumed hours that could have been spent otherwise. That expenditure is the message; the likeness is only the medium. Generative tools sever effort from output. What remains looks like the artefact of devotion whilst carrying none of the expenditure that made it legible, so the form survives and the signal dies, and senders compensate with volume, which is precisely the behaviour that produces the flood.

Parasocial grief research cuts against the easy dismissal too. A 2023 experiment by Alysse C. Baker and Elizabeth L. Cohen, published in Death Studies, ran 271 participants through imagined deaths of close and distant figures, both acquaintances and celebrities, and found that relational closeness, not parasociality, predicted the intensity of grief. People who felt close to a celebrity anticipated grief comparable to that for a close friend. Whether the relationship was one-sided did not matter. The fans posting these images are not confused about what they feel. Their grief is real and it is proportionate to how close they felt. Their error is a category error about audience: they are performing a private feeling in a public space the family also inhabits, and where the family cannot refuse delivery.

The Money That Keeps the Feed Full

Sincerity has a free rider. A 2026 study by the Centre for Media, Technology and Democracy mapped 340 Facebook pages, operated principally from Vietnam, which published 355,296 posts across 2025 and 2026, peaking at 3,919 in a single day in May 2026. The pages had a median following of around 11,000 and harvested more than 95 million reactions, comments and shares. Of 5,894 posts aimed at Canadian audiences, 54 per cent were false or unsupported, typically pairing a genuine topic with a fabricated quote, statistic or invented dramatic scene. Revenue came through ordinary advertising exchanges, including Google's, so legitimate organisations were unknowingly funding the operation.

Celebrity death is the highest-yielding raw material this industry has. Emotion is pre-loaded, search volume spikes without anyone having to manufacture interest, the subject cannot issue a correction, and the audience arrives disposed to share. The pattern was documented after the killing of Charlie Kirk in September 2025, when Facebook filled with AI-generated images of celebrities making invented heartfelt announcements, each post functioning as a small cash machine. Fact-checkers who tried to debunk them reported there were simply too many to cover.

The platforms are not passive here. The Hype feature that pushed the fake Parton thumbnails onto the Videos on the Rise shelf was built to boost discovery for smaller channels, and it did exactly that. YouTube told Mashable it was reviewing the content and would act against anything violating its policies, a sentence platforms have issued since roughly 2011, describing a process that runs at human speed against a supply that runs at machine speed.

In November 2025 Giovanni Spitale and Federico Germani posted a paper called “The Making of Digital Ghosts: Designing Ethical AI Afterlives,” later published in Ethics and Information Technology. It is the most systematic attempt yet to say what an acceptable simulation of a dead person would have to look like. The authors build a nine-dimensional taxonomy covering timing, consent, data sources, interaction modality, fidelity and disclosure, purpose, audience and access, governance and ownership, and autonomy. Each dimension carries independent moral weight, which is why blanket approval and blanket prohibition both fail.

Their conclusion is stringent. An ethically acceptable digital ghost should be rooted in premortem intent, created because the person decided during life that they wanted it rather than assembled afterwards by survivors or strangers. It requires mutual consent, transparent and limited data use, clear disclosure, restricted purposes and access, stewardship by family or estate, and minimal behavioural agency, meaning it should not act on its own initiative.

Apply that to the Parton flood and it fails on every dimension at once. There was no premortem intent; there was a recorded premortem preference pointing the other way. There was no mutual consent, no disclosure worth the name beyond an automated label under a fabricated deathbed, no restriction of purpose or access, no stewardship. The framework is a good one, and it was written for a world in which digital afterlives are products, built by companies, for customers, under contracts. The Parton case is a world in which the afterlife is made by unaffiliated individuals at no cost, with no contract to attach conditions to.

The commercial world is arriving regardless. In December 2025 Meta was granted a patent, filed in 2023, describing a language model that could post, comment, send voice messages and simulate calls on behalf of dormant or deceased account holders. Meta said it had no plans to move forward with that example. The cyberpsychologist Elaine Kasket, who has studied digital afterlives for two decades, read it as treating user death as an engagement problem, a very uncomfortable and not very psychologically helpful turn. Sherry Turkle of MIT argued it interrupts grieving and invites the bereaved to deny death in favour of a perpetual fantasy life.

A Voice Is Harder to Own Than a Face

Among the artefacts circulating after Parton's death, the songs are legally the strangest. In June 2026 Jyh-An Lee and Xuan Sun of the Chinese University of Hong Kong posted “Vocal Identity Under Siege by AI Voice Cloning Technologies,” later published in the Singapore Journal of Legal Studies. Prompted partly by the resemblance between OpenAI's ChatGPT-4o voice and that of Scarlett Johansson, they compare three doctrinal routes to protecting a voice: the American right of publicity, continental personality rights, and personal data protection. Each has distinct gaps, and they pay particular attention to posthumous protection, where the divergence between traditions is widest.

Tennessee is the relevant jurisdiction for a Nashville estate, and it is unusually well equipped. The ELVIS Act, signed in March 2024 and effective from 1 July, amended the state's 1984 Personal Rights Protection Act to add voice, defined to include a simulated voice, to the protected attributes of name, photograph and likeness. It went further than any prior American statute by attaching liability to anyone who makes available a tool, service or device whose primary purpose is to produce a particular identifiable individual's photograph, voice or likeness without authorisation. Tennessee's post-mortem right runs for ten years after death and can continue indefinitely so long as it is commercially exploited, lapsing only after a two-year gap in use. On paper, Parton's estate has as strong a position as any estate anywhere.

On paper is doing a lot of work there. The ELVIS Act carries a fair use exception covering news and public affairs, comment, criticism, scholarship, satire, parody and fleeting or incidental uses. A tribute song posted without advertising, or a memorial image shared to a personal account, has a plausible claim to sit inside those carve-outs, and the statute's centre of gravity is unauthorised commercial use rather than uncompensated devotion.

Federal law may arrive. The NO FAKES Act of 2026, S.4591, cleared the Senate Judiciary Committee on a unanimous vote on 18 June 2026 with fifteen cosponsors from both parties. It would create a federal property right in voice and visual likeness for every individual, celebrity or not, and impose platform liability for hosting unauthorised replicas with knowledge of non-consent, policed through DMCA-style notice and takedown with a $25,000 penalty for false counter-notification. It excludes First Amendment protected speech including news reporting, parody and criticism. As of September 2026 it has not reached the Senate floor, and trackers put its odds of enactment at around one in four.

Europe has taken two swings. Article 50(4) of the EU AI Act requires deployers of systems generating deepfake image, audio or video content to disclose that the content is artificially generated, an obligation applying from 2 August 2026, with transitional relief for the machine-readable marking duty in Article 50(2) running to December 2026 for systems already on the market. Denmark went further and stranger, unveiling on 26 June 2025 a proposed Copyright Act amendment granting individuals a neighbouring right over their own body, facial features and voice, enforceable by takedown and compensation without proof of reputational harm, with carve-outs for parody, satire, caricature and social criticism. The draft carries a fifty-year post-mortem term, but the intended commencement of 31 March 2026 passed without the amendment entering into force, a snap general election having broken the parliamentary timetable; revised targets conflict, variously 1 July 2026 or, if it is adopted at all, the third quarter. The European Commission, notified of the amended bill under the TRIS procedure on 31 October 2025, replied on 19 March 2026 with objections that are concrete rather than merely sceptical: copyright protects original works reflecting an author's own intellectual creation, and a face or a voice is not one; no single member state can unilaterally place a digital imitation of a voice inside copyright's scope without unpicking EU harmonisation; and the proposed upload-filter obligations exceed what the framework permits, built as such tools are to detect copyright-protected content rather than to monitor a person's appearance or voice. The concern about deepfakes is legitimate, the Commission allowed, but the remedy should be EU-wide. Lawyers now read the route of granting a person copyright in their own face as, for the present, closed off, and what replaces it is genuinely unclear. Anyone telling you otherwise is guessing.

Where Every Proposed Remedy Stops

Here is the load-bearing problem with all of it.

The AI Act's Article 3(4) defines a deployer as a natural or legal person using an AI system under their authority, “except where the AI system is used in the course of a personal non-professional activity.” A grieving fan in Rotterdam generating a picture of Dolly Parton among angels and posting it to their own account is not a deployer. Europe's flagship synthetic media transparency regime, in its most directly applicable provision, exempts precisely the person who produces most of this material.

The Tennessee statute reaches commercial exploitation and tool distribution rather than personal posting, and exempts comment and satire besides. NO FAKES, if it passes, hinges on notice and takedown, which requires the estate to find each item, notify each platform and wait. Denmark's approach faces the enforcement difficulty its own advocates concede, namely that most infringements originate outside Danish jurisdiction.

Platform tooling has the same shape. YouTube's likeness detection, previewed in 2024, rolled out through the Partner Programme in late 2025 and opened to all creators aged eighteen and over in 2026, is a genuinely useful instrument. To enrol, you open the Likeness tab under Content Detection in YouTube Studio, scan a QR code, submit a government ID and record a selfie video. Consider what that asks of a dead woman. The most advanced likeness protection any platform has shipped is gated behind a liveness check, and the people who most need it cannot pass it.

Platform discretion is no better. When OpenAI launched Sora with a ban on generating videos of public figures, the ban did not cover dead ones, and the app filled with Michael Jackson, Bob Ross and Tupac Shakur. Zelda Williams asked people to stop sending her AI videos of her father. In October 2025 OpenAI blocked Martin Luther King Jr after his estate complained about disrespectful depictions, and said it would remove recently deceased public figures where families or estates asked. That is a remedy for estates with lawyers, granted at a company's discretion, after the fact.

The pattern is consistent. Every mechanism on offer regulates somebody who can be identified, served, sued or deplatformed. None reaches the person with a phone and a free image generator, acting from what they believe is love, posting to an audience of two hundred. Open-weight models mean that even a total policy victory over every commercial provider would leave the underlying capability sitting on consumer hardware, permanently.

What Might Actually Reduce the Volume

Three interventions look worth the effort, and each has an obvious ceiling.

Provenance at capture rather than labels at distribution. The MIT finding that disclosure failed to prevent false memory formation is the most operationally important result in that paper. A badge under a fabricated hospital photograph did not stop people remembering the hospital. Cryptographic provenance attached to genuine material might work better, letting the real archive of a life be positively authenticated rather than the synthetic flood negatively flagged. The ceiling: provenance metadata is trivially stripped, and most of the Parton archive predates any such standard by decades.

A bereavement window in ranking systems. Platforms already run elevated-risk protocols around elections and breaking violence. A defined period after a verified death during which synthetic depictions of the deceased are excluded from algorithmic amplification, rather than removed, would have kept a fabricated deathbed image off Videos on the Rise on 27 August. Speech stays intact; the distribution subsidy goes. The ceiling: it treats the ten days that most damage a family and does nothing about the years afterwards, when the artefacts sit in the index.

Estate enrolment that does not require a pulse. If likeness detection is going to exist, the verification path has to accommodate executors acting for the dead, with documentary proof of authority substituting for a selfie. This is an unglamorous engineering and legal problem rather than a philosophical one, which is probably why nobody has solved it.

None of these touch the sincere fan. Nothing will. That is not defeatism, it is the constraint the problem actually has, and pretending otherwise produces the sort of policy-shaped ending that makes readers feel resolved and changes nothing.

What remains is the cultural question Stella Parton was actually asking. Her plea was not for legislation. She asked people to show more compassion and wisdom, and to treat her sister's life as an example for tolerance and respect. Stripped of sentiment, that is a request to recognise that the bereaved are a real audience for content addressed to the dead, and that they cannot opt out.

There is an older answer, and it costs nothing. Dolly Parton spent sixty years being photographed. The genuine archive is enormous, freely available, and depicts a woman who actually existed, doing things she actually did, wearing clothes she chose. Every one of those images is a more faithful tribute than any generation, for the simple reason that she was there when it was made. The reflex that reached for a prompt instead is not wickedness. It is a habit, acquired over perhaps three years, of treating generation as the default mode of expression. Habits can be unlearned faster than statutes can be passed.

The memory researchers would add one thing. On the evidence of the MIT study, images we make of the dead do not simply accumulate alongside the record. They compete with it, they win a measurable share of the time, and the people who fall for them grow more confident as they grow less accurate. A fabricated photograph of Dolly Parton in a hospital bed is not only false. It is, for some proportion of those who see it, the way she will be remembered as having died. Her sister, who was actually there, will spend the rest of her life scrolling past it.

Sources and References

  1. NPR. (2026, August 25). “Dolly Parton, one of country music's most singular performers, has died at age 80.” Retrieved from https://www.npr.org/2026/08/25/nx-s1-3650175/dolly-parton-obituary
  2. NBC News. (2026, September 5). “Dolly Parton's sister slams 'fake AI garbage' flooding the internet since icon's death.” Retrieved from https://www.nbcnews.com/pop-culture/pop-culture-news/dolly-partons-sister-slams-fake-ai-garbage-rcna596080
  3. TODAY. (2026, September 5). “Dolly Parton's Sister Slams 'AI And Endless Garbage' Amid Singer's Death.” Retrieved from https://www.today.com/popculture/news/dolly-parton-sister-stella-slams-ai-tributes-rcna596226
  4. HuffPost. (2026, September 5). “Dolly Parton's Sister Criticizes AI Posts After Trump Tribute.” Retrieved from https://www.huffpost.com/entry/dolly-parton-sister-stella-ai-garbage-trump_n_6a9d4f35e4b0783b3d92fe3a
  5. Mashable, via Yahoo News. (2026, August 27). “YouTube creators are using fake Dolly Parton images.” Retrieved from https://www.yahoo.com/news/us/articles/youtube-creators-using-fake-dolly-211139253.html
  6. Futurism. (2026, September). “Dolly Parton's Sister Begs Public to Stop Posting 'Fake AI Garbage' in the Wake of the Beloved Singer's Death.” Retrieved from https://futurism.com/artificial-intelligence/dolly-parton-sister-slop
  7. E! News. (2025, October 8). “Dolly Parton Slams AI Photo of Reba McEntire at Her 'Deathbed'.” Retrieved from https://www.eonline.com/news/1423576/dolly-parton-on-ai-deathbed-reba-mcentire-photo
  8. Deadline. (2023, July 4). “Dolly Parton Weighs In On Becoming An AI Hologram After She Dies: 'I Don't Want To Leave My Soul On Earth'.” Retrieved from https://deadline.com/2023/07/dolly-parton-ai-hologram-after-death-1235429779/
  9. Pataranutaporn, Pat, Chayapatr Archiwaranguprok, Samantha W. T. Chan, Elizabeth Loftus and Pattie Maes. (2024, September 13). “Synthetic Human Memories: AI-Edited Images and Videos Can Implant False Memories and Distort Recollection.” arXiv preprint 2409.08895; published in Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. Retrieved from https://arxiv.org/abs/2409.08895
  10. Sacchi, Dario L. M., Franca Agnoli and Elizabeth F. Loftus. (2007). “Changing history: doctored photographs affect memory for past public events.” Applied Cognitive Psychology, 21(8). Retrieved from https://onlinelibrary.wiley.com/doi/10.1002/acp.1394
  11. Spitale, Giovanni and Federico Germani. (2025, November 25). “The Making of Digital Ghosts: Designing Ethical AI Afterlives.” arXiv preprint 2511.20094; published in Ethics and Information Technology. Retrieved from https://arxiv.org/abs/2511.20094
  12. Lee, Jyh-An and Xuan Sun. (2026, June 11). “Vocal Identity Under Siege by AI Voice Cloning Technologies.” arXiv preprint 2606.12812; Singapore Journal of Legal Studies. Retrieved from https://arxiv.org/abs/2606.12812
  13. Li, Yifan and Xingyu Lan. (2026, March 1). “Remember You: Understanding How Users Use Deadbots to Reconstruct Memories of the Deceased.” arXiv preprint 2603.01017. Retrieved from https://arxiv.org/abs/2603.01017
  14. Baker, Alysse C. and Elizabeth L. Cohen. (2023, November 2). “Relational closeness, not parasociality, determines the intensity of grief responses to celebrity death.” Death Studies, 48(8). Retrieved from https://www.tandfonline.com/doi/full/10.1080/07481187.2023.2276301
  15. Klass, Dennis, Phyllis R. Silverman and Steven L. Nickman, eds. (1996). Continuing Bonds: New Understandings of Grief. Routledge. Retrieved from https://www.routledge.com/Continuing-Bonds-New-Understandings-of-Grief/Klass-Silverman-Nickman/p/book/9781560323396
  16. NPR. (2025, October 17). “OpenAI blocks MLK Jr. videos on Sora after 'disrespectful depictions'.” Retrieved from https://www.npr.org/2025/10/17/nx-s1-5577869/sora-block-videos-mlk
  17. Engadget. (2026). “YouTube's AI deepfake detection tool is now available to all creators 18 and older.” Retrieved from https://www.engadget.com/2174282/youtube-likeness-detection-ai-deepfakes-expansion/
  18. Fortune. (2026, March 3). “Meta patented an AI model that could take over a dead person's profile.” Retrieved from https://fortune.com/2026/03/03/meta-patent-ai-model-death-profile-commenting-psychology-grief/
  19. Holland & Knight. (2024, April). “First-of-Its-Kind AI Law Addresses Deep Fakes and Voice Clones.” Retrieved from https://www.hklaw.com/en/insights/publications/2024/04/first-of-its-kind-ai-law-addresses-deep-fakes-and-voice-clones
  20. Rothman's Roadmap to the Right of Publicity. “Tennessee.” Retrieved from https://rightofpublicityroadmap.com/state_page/tennessee/
  21. Congress.gov. (2026). “S.4591, NO FAKES Act of 2026, 119th Congress.” Retrieved from https://www.congress.gov/bill/119th-congress/senate-bill/4591
  22. Holland & Knight. (2026, June). “Senate Committee Advances Bill to Protect Name, Image, Likeness and Voice Against Unauthorized AI Use.” Retrieved from https://www.hklaw.com/en/insights/publications/2026/06/senate-judiciary-committee-advances-legislation-to-protect-name
  23. EU Artificial Intelligence Act. “Article 50: Transparency Obligations for Providers and Deployers of Certain AI Systems.” Regulation (EU) 2024/1689. Retrieved from https://artificialintelligenceact.eu/article/50/
  24. Schjodt. (2025). “Owning the Self: Denmark's Copyright Turn Against Deepfakes.” Retrieved from https://schjodt.com/news/owning-the-self-denmarks-copyright-turn-against-deepfakes; ICTRecht. (2026). “The EC pushes back on Denmark: no copyright on your own face?” Retrieved from https://www.ictrecht.nl/en/blog/de-ec-fluit-denemarken-terug-toch-geen-copyright-op-je-eigen-gezicht
  25. Centre for Media, Technology and Democracy. (2026). “Facebook's News Slop Economy: Inside a foreign-run AI content-farm network monetizing fake news across Canada, the US, the UK, and Europe.” Retrieved from https://mediatechdemocracy.com/en/publications/facebook-news-slop-economy_2026/

Tim Green

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...