The American clothing brand J. Crew is under fire after it was revealed that the company used AI to guzzle up its own aesthetic and promote ads with seemingly fake models.
The images, which were published to Instagram earlier this month, don't raise immediate alarm bells. Presumed human men are pictured embodying J. Crew's vintage Americana-prep aesthetic: vibing on boats, cycling past storefronts in muted colors, drawing and painting in their paint-splattered studio, and so on.
But as style blog <a href="https://www.blackbirdspyplane.com/p/jcrew-used-ai-to-counterfeit-their-own-vibes">Blackbird Spyplane first spotted</a>, upon closer inspection the images are absolutely littered with signs of AI slop. The errors range from small oddities — misaligned stripes on a rugby shirt, for instance — to more blatant errors, like one man's foot facing backwards and another's hands melting into the handlebars of his bike.
<blockquote class="tiktok-embed" cite="https://www.tiktok.com/@edgyalbert/video/7543021646144097566" data-video-id="7543021646144097566" data-embed-from="oembed" style="max-width:605px; min-width:325px;">
<section> <a target="_blank" title="@edgyalbert" href="https://www.tiktok.com/@edgyalbert?refer=embed">@edgyalbert</a> 
AI SLOP X J CREW
 <a target="_blank" title="♬ original sound - Albert Muzquiz" href="https://www.tiktok.com/music/original-sound-7543021642193160991?refer=embed">♬ original sound - Albert Muzquiz</a> </section>
</blockquote>
 <script async src="https://www.tiktok.com/embed.js"></script>
After Blackbird Spylane's reporting unmasked the slop, the company updated its Instagram captions to include a callout to someone named Sam "AI Sam" Finn, who refers to themselves as an "AI photographer." Still, J. Crew didn't actually cop to the use of AI in the caption; instead, they referred to the campaign images as "digital art." The captions also don't acknowledge whether the depicted models are real people or not.
As <a href="https://www.thecut.com/article/did-jcrew-use-ai-on-instagram-campaign.html?isNewSocialUser=false&amp;providerId=google.com">The Cut pointed out</a>, there were a large number of folks in the comments section who expressed annoyance and anger at the images. Which makes sense! Unbeknownst to J. Crew, it seems, a <a href="https://futurism.com/vogue-magazine-ai-model-ad">lot of people do care</a> whether the humans used in the advertisements designed to sell them clothes exist — and, to that end, whether the actual items J. Crew is actively trying to sell them are shown in their real form, or are a hyperreal, AI-generated version.
<blockquote class="twitter-tweet" data-width="500" data-dnt="true">
Those J.Crew images from the other day that evoked their old ‘80s and ‘90s catalogs are in fact A.I. generated.
A Spyplane Exclusive Investigation, out now<a href="https://t.co/LhIIsdpUzg">https://t.co/LhIIsdpUzg</a>
&mdash; Blackbird Spyplane (@BLKBRD_Spyplane) <a href="https://twitter.com/BLKBRD_Spyplane/status/1960350157662527575?ref_src=twsrc%5Etfw">August 26, 2025</a></blockquote>
<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
When The Cut reached out for comment, J. Crew issued a terse response — that, once again, failed to acknowledge the use of AI.
"We're always exploring new forms of creative expression, expanding how we work, and finding fresh, innovative ways to create content," the company told the magazine. "This partnership with Sam Finn Studio is one of many examples of how we engage artists of all genres to interpret our brand and experiment with different art mediums."
We don't know what's sadder: J. Crew's lack of disclosure around AI use, its decision to leave sloppy AI errors all over the campaign, or its belief that its iconic aesthetic, born out of decades of design and creative human labor, should be swallowed up and ralphed back out by AI.
More on AI slop: <a href="https://futurism.com/wired-business-insider-ai-articles">Wired and Business Insider Accidentally Published AI-Generated Slop Articles by Seemingly Fake Journalist</a>

Slopwatch: Fashion Edition

Just... why?

J. Crew Is in Hot Water After Its “Vintage” Ad Turned Out to Be Faked Using AI

In the nearly 13 years since Tinder launched, the dating app has been everything from a <a href="https://www.theguardian.com/lifeandstyle/2016/may/22/hook-up-apps-grindr-tinder-gay-straight-people-dating">straight alternative to Grindr</a> to a <a href="https://theconversation.com/tinder-fails-to-protect-women-from-abuse-but-when-we-brush-off-dick-pics-as-a-laugh-so-do-we-147909">sexual harassment factory</a> to a venue to <a href="https://www.businessinsider.com/husband-wife-met-tinder-during-hookup-eras-2024-9">meet one's spouse</a> — and, most recently, an <a href="https://www.bloomberg.com/news/newsletters/2024-02-14/scammers-litter-dating-apps-with-ai-generated-profile-pics?sref=YfHlo0rL">AI-riddled graveyard</a>.
Nevertheless, there are still flesh-and-blood humans successfully using the app. And according to a fascinating new study about the folks who successfully hooking up on it, they share some bizarre traits in common.
In a <a href="https://www.liebertpub.com/doi/10.1089/cyber.2023.0256">new paper</a> published in the journal Cyberpsychology, Behavior, and Social Networking, psychology researchers found that men who reported having a lot of sex with people they met on Tinder tended show signs of, in a word, psychopathy — and their female counterparts, unfortunately, just seemed to have lower standards.
Using a sample of nearly 500 German-speaking adults from the wide age range of 16 to 70 years old, personality psychologists Lennart Freyth of Johannes Kepler University and Peter K. Jonason the University of Cardinal Stefan Wyszyński gave their subjects — all recruited online and given a small monetary incentive — a battery of questions assessing their personality traits, experiences with Tinder and similar dating apps, as well as their sexual and romantic decision-making processes.
As with <a href="https://www.psychologytoday.com/us/blog/the-mysteries-love/202106/new-study-investigates-dark-triad-traits-and-link-promiscuity">other studies</a> relating to <a href="https://www.tandfonline.com/doi/full/10.1080/14681994.2025.2507789">sociosexual psychology</a>, the study's participants were assessed for "Dark Triad" personality traits, which include narcissism (<a href="https://www.psychologytoday.com/us/basics/narcissism">defined by Psychology Today</a> as an overblown sense of self-importance), Machiavellianism (cunningness and manipulation, <a href="https://www.tandfonline.com/doi/full/10.1080/00223980.2024.2382243i">per the</a><a href="https://www.tandfonline.com/doi/full/10.1080/00223980.2024.2382243i"> Journal of Psychology</a>), and the <a href="https://psychopathysociety.org/page/AboutPsychopathy">often-misunderstood</a> condition of psychopathy, which the <a href="https://www.apa.org/monitor/2022/03/ce-corner-psychopathy">American Psychological Association characterizes</a> as involving "low empathy and remorse... impulsivity, and sometimes aggressive or violent behavior."
Additionally, the survey respondents were asked questions their feelings and experiences surrounding sex and dating, including their relationships to casual sex, the open expression of sexual desire, and their past sexual experiences. They were also asked about how satisfied they felt with their potential matches on dating apps, and how often they look around for new ones.
Finally, the participants were asked whether they had ever been on date with someone from Tinder or a similar app, and whether they had had sex with someone from said apps.
As the psychology experts found, roughly 31 percent of the participants, male and female alike, said they had gone on a Tinder date. Of the participants who reported having had sex with someone from a dating app — the percentage of which was not, unfortunately, listed — men were way more likely than women to do so, with 66 percent of that cohort being male.
Cross-referencing with the Dark Triad assessments, the researchers concluded that the men who had sex with people from the apps were more likely to score higher in unrestricted or openly-discussed sexual desire — no surprises there — but also, troublingly, in psychopathy. The same couldn't be said of men who didn't report having slept with app matches, nor of any women in the study.
The women who said they'd had Tinder sex, meanwhile, reported much higher satisfaction with their decisions than the women who did not. The personality psychologists didn't determine exactly why those women felt so gratified by their app exploits, but they suggested that those women may have been more willing to accept flaws or mismatches in person, or that they felt an increased sense of sexual agency in doing so.
Speaking to PsyPost <a href="https://www.psypost.org/psychopathic-men-and-less-selective-women-report-more-sex-via-tinder/">about the study</a>, Freyth drove home his surprise at learning that the women in his survey seemed to be "lowering their standards once on a date."
"We consider this something like a sunk cost effect — after already texting, once on a date, the decision to engage further was already made," he said.
(Interestingly, it appears that Freyth and Jonason were also not surprised to find that men who hook up on dating apps <a href="https://www.psychologytoday.com/us/blog/why-bad-looks-good/202008/which-dating-apps-appeal-to-dark-personalities">tend to be more psychopathic</a>.)
The paper's authors acknowledge that their study has its limitations, including that its self-reported responses could be skewed due to shame or self-aggrandizing. Still, Freyth boasted that the survey's age range made it more comprehensive than <a href="https://www.sciencedirect.com/science/article/pii/S0736585323000138">others</a> which have <a href="https://www.tandfonline.com/doi/full/10.1080/02673843.2023.2241524">focused only</a> on <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC11110432/">younger people</a>.
"Compared to most studies on dating apps," he told PsyPost, "our sample was relatively old (the average age was around 40 years)."
There's also the geographical component: how different would a study looking into the dating app experiences of 16-to-70-year-olds be if it were conducted in East Asia, or Latin America, or the United States? For now, we can only speculate — but we'd bet that there would be some fascinating takeaways for each.
More on sex: <a href="https://futurism.com/neoscope/scientists-identify-sex-range">Scientists Identify Ideal Amount of Boinking for Maximum Mental Health</a>

Get Some Strange

Studies

Fascinating -- and dark.

Scientists Have Figured Which Types of People Are Successfully Hooking Up the Most on Tinder

Tesla's misleadingly-named "<a href="https://futurism.com/judge-tesla-failed-full-self-driving-refund">full-self driving</a>" (FSD) feature — which still requires drivers pay attention to the road, even if they often <a href="https://futurism.com/apple-vr-driving-tesla">fail horrifically</a> at that <a href="https://futurism.com/the-byte/tesla-911-motorcyclist-fsd">simple task</a> — is one of the Elon Musk-owned electric carmaker's biggest <a href="https://www.reuters.com/business/autos-transportation/elon-musk-pins-hopes-full-self-driving-teslas-next-profit-driver-2023-04-20/">self-styled selling points</a>.
Until recently, the boastful multi-hyphenate entrepreneur's <a href="https://futurism.com/video-elon-musk-promising-self-driving-cars">decade-plus of over-promising</a> and <a href="https://futurism.com/elon-musk-realizes-all-teslas-self-driving-computers">under-delivering</a> on assisted driving software did little to tarnish <a href="https://insideevs.com/news/581046/ark-invest-tesla-share-prediction-includes-robotaxis/">bullish investors</a>' slavish devotion to Tesla. But now, as a <a href="https://ev-intelligence.com/evs-self-driving-cars-aug2025/">new survey</a> from the consulting firm Slingshot Strategies found, it looks like the FSD bubble has finally burst, even as <a href="https://futurism.com/tesla-dismal-sales-news-robotaxi-launch">Tesla's abysmal sales</a> and its CEO's <a href="https://futurism.com/elon-musk-ponders-unpopularity">ever-toxic politics</a> threaten to drive the company <a href="https://futurism.com/elon-musk-realizes-tesla">straight into the ground</a>.
In its monthly Electronic Vehicle Intelligence Report (EVIR), Slingshot surveyed more than 8,000 people across US demographics in August, finding that 35 percent of participants said having FSD included in an EV they were considering for purchase would now make them less likely to buy it. By contrast, only 14 percent said that Tesla's self-driving suite, which costs either <a href="https://www.teslarati.com/tesla-overhauling-full-self-driving-subscription-easier-access/">$8,000 upfront</a> or $99 per month, would convince them to purchase a car.
(The final 51 percent couldn't care less either way, interestingly, lending credence to Musk's acknowledgement in a <a href="https://www.fool.com/earnings/call-transcripts/2025/07/23/tesla-tsla-q2-2025-earnings-call-transcript/">July earnings call</a> that the "vast majority of people don’t know [FSD] exists.)
Another fascinating data point: nearly half of the survey respondents said they think the tech — which along with Tesla's Autopilot feature has been linked to <a href="https://futurism.com/the-byte/tesla-autopilot-fsd-linked-crashes">hundreds of crashes</a>, some fatal — shouldn't be legal at all.
The not-so-self-driving feature has faced ample regulatory scrutiny over safety concerns and its maddeningly <a href="https://futurism.com/the-byte/yann-lecun-elon-musk-lying-fsd">misleading brand name</a>.
As <a href="https://www.telegraph.co.uk/business/2025/03/08/britain-frustrates-musk-blocking-launch-self-driving-tesla/">The Telegraph reported</a> earlier this year, the United Kingdom decided against letting motorists take their hands off the wheel when self-driving features were engaged. In China, as <a href="https://www.reuters.com/business/autos-transportation/tesla-says-will-release-fsd-feature-china-after-software-approval-2025-03-24/">Reuters reported in March</a>, a one-month free trial of <a href="https://futurism.com/tesla-suspends-full-self-driving-china">FSD was abruptly suspended</a> pending regulatory approval — and days after, the feature's name was changed, without fanfare, to "Intelligent Assisted Driving."
Though the anti-FSD EVIR respondents didn't provide any reasons why they oppose the feature, some on social offered their own takes.
"Frankly, I don't see how FSD is legal in the first place," Philadelphia-based automotive liability attorney <a href="https://bsky.app/profile/maxkennerly.bsky.social/post/3lximly2sl22p">Max Kennerly posted</a>. "It's not certified or approved by anyone."
Author Edward Niedermeyer, whose book "<a href="https://bookshop.org/p/books/ludicrous-the-unvarnished-story-of-tesla-motors-edward-niedermeyer/16657578">Ludicrous: The Unvarnished Story of Tesla Motors</a>" massively <a href="https://www.cybertruckownersclub.com/forum/threads/manufactured-hate-you-need-to-know-who%E2%80%99s-behind-this.38490/">peeved Musk's rabid fanboys</a> when it was released in 2019, took aim at the automaker's financial incentives.
"The problem isn't consumers, the problem is investors who think Tesla is on the brink of 'solving self-driving' and bid up the stock to a trillion dollar valuation," <a href="https://bsky.app/profile/niedermeyer.online/post/3lximbkwjo22a">Neidermeyer posted</a>.
Ever-present at any scene of any FSD criticism, professional Musk-hater Dan O'Dowd, the billionaire founder of the nonprofit Dawn Project that seeks to <a href="https://dawnproject.com/banfsd/">shut down</a> the feature, suggested in his <a href="https://bsky.app/profile/realdanodowd.bsky.social/post/3lxixnybnfc2e">own post</a> that that "consumers are smart, [and] they do not trust Tesla."
More on not-so-self-driving: <a href="https://futurism.com/video-tesla-robotaxi-safety-monitor-driving">Video Shows Tesla Robotaxi Safety Driver Giving Up, Climbing Into Driver's Seat</a>

Liability Insurance

"Frankly, I don't see how FSD is legal in the first place."

Survey Finds That Self-Driving Is Actively Chasing Buyers Away From Tesla

Advocates have long contended that so-called "<a href="https://futurism.com/halo-people-mad-ai-glasses-record-everything">smart glasses</a>," which when activated record whatever the wearer sees, pose serious privacy risks.
There was a <a href="https://www.wired.com/2013/12/glasshole/">sharp outcry</a> when Google debuted Project Glass more than a decade ago, and now with Meta's Ray-Ban AI glasses, a new generation is finding out the hard way just how creepy these surveillance wearables can be.
In a <a href="https://www.tiktok.com/@niessaxoxo_/video/7538155695066074398">recent viral TikTok video</a>, influencer Aniessa Navarro said that she was in the middle of a Brazilian wax — which involves removal of body hair from intimate areas — at her regular Manhattan location of the European Wax Center chain when she realized that her esthetician was wearing a pair of Mark Zuckerberg's spy spectacles.
Surprised, Navarro asked the esthetician if she was "wearing Meta glasses," and the EWC employee copped to it — but promised they were neither charged nor turned on, and claimed they were prescription as well.
"After that I kind of shut down, and I could not stop thinking, 'could this girl be filming me right now?'" the influencer recounted. "I literally could not stop thinking about it the whole entire time."
Visibly perturbed, Navarro asked her followers for advice on what to do, and many urged her to contact lawyers and EWC headquarters. In a <a href="https://www.tiktok.com/@niessaxoxo_/video/7538588811941448990">followup</a>, the young woman said that she emailed the waxing chain and was given only a "generic" response — and when EWC sent her a customer satisfaction survey, she again expressed her distress at the situation.
Once the incident started attracting media attention, the chain had more to say. In a <a href="https://www.washingtonpost.com/technology/2025/08/27/gen-z-smart-glasses-digital-privacy-consent/">statement to the Washington Post</a> about the upsetting incident, a spokesperson for the company provided a bit more of a response, claiming without evidence that the employee's smart glasses had been powered down while she was performing the procedure.
Regardless, Navarro said in her second video that she had already spoken to two law firms, which both told her she could well have a case.
Still, she appeared to get cold feet because she didn't want to "get anyone fired if she did not do anything wrong," and is more interested in raising awareness and getting those sorts of recording devices banned in such intimate settings.
Unfortunately, this is not the first such incident of inappropriate recording using Meta's $350 AI Wayfarers. After an exotic dancer <a href="https://www.reddit.com/r/stripper/comments/1g55u1s/ladies_beware_psa/">warned last year on Reddit</a> that she'd seen an uptick of "creep men wearing [Ray-Ban] Meta glasses in the club," a female bouncer at a gentleman's club in Atlanta <a href="https://www.tiktok.com/@britknee0405/video/7527471434466217271">recorded herself</a> last month having a tense exchange with a man who tried to wear them in her establishment.
"I wear 'em everywhere!" the off-screen man intones.
"Not here!" the bouncer responded. When the patron asked why he couldn't, the bouncer noted plainly: "because they've got cameras in them."
Ultimately, the unseen club patron complies and leaves his recording glasses at the door — but his reticence to do so, and even his entitlement at wearing them to the club in the first place, says a lot about what kind of future we might be in for once those sorts of surveillance devices become cheaper and more commonplace.
More on smart glasses: <a href="https://futurism.com/smart-glasses-vibe-thinking">Harvard Startup Says Its Smart Glasses Will Do "Vibe Thinking" for You</a>

Panopti-Glasses

"I could not stop thinking, 'could this girl be filming me right now?'"

Woman Goes to Get Brazilian Wax, Alarmed to Notice Waxer Is Wearing Meta’s Video Recording Glasses

A single typo, formatting error, or slang word makes an AI more likely to tell a patient they're not sick or don't need to seek medical care.
That's what MIT researchers found in a <a href="https://dl.acm.org/doi/10.1145/3715275.3732121">June study</a> currently awaiting peer review, which we <a href="https://futurism.com/ai-something-bizarre-typos">covered previously</a>. Even the presence of colorful or emotional language, they discovered, was enough to throw off the AI's medical advice.
Now, in a <a href="https://www.bostonglobe.com/2025/08/27/business/mit-ai-medical-errors-bias/">new interview</a> with the Boston Globe, study coauthor Marzyeh Ghassemi is warning about the serious harm this could cause if doctors come to widely rely on the AI tech.
"I love developing AI systems," Ghassemi, a professor of electrical engineering and computer science at MIT, told the newspaper. "But it's clear to me that naïve deployments of these systems, that do not recognize the baggage that human data comes with, will lead to harm."
AI could end up causing discrimination against patients who can't communicate clearly in English, native speakers with imperfect command of the language, or anyone that may commit the human mistake of speaking about their health problems emotionally. Doctors using an AI tool could feed it their patients' complaints that were sent over an email, for example, raising the risk that the AI would give them bad advice if those communications weren't flawlessly composed.
In the study, the researchers pooled together patients' complaints taken from real medical records and health inquiries made by users on Reddit. They then went in and dirtied up the documents — without actually changing the substance of what was being said — with typos, extra spaces between words, and non-standard grammar, like typing in all lower case. But they also added in the kind of unsure language you'd expect a patient to use, like "kind of" and "possibly." They also introduced colorful turns of phrase, like "I thought I was going to die."
From there, they fed these cases to four different AI models, including OpenAI's GPT-4 — though, to be fair, none that were particularly cutting-edge — to judge if a patient should visit their doctor, get lab work done, or not come in at all. The numbers were striking: overall, the AI tools were seven to nine percent more likely to recommend patients not to seek medical care at all when reading complaints with imperfect — but arguably more realistic — language.
"Adding additional information, even if true and relevant, often reduced the accuracy of models," Paul Hager, a researcher at the Technical University of Munich who was not involved in the study, told the Globe. "This is a complex issue that I think is being addressed to some degree through more advanced reasoning models... but there is little research on how to solve it on a more fundamental level."
That the bots are woefully inaccurate isn't surprising. Hallucinations, instances of a chatbot <a href="https://futurism.com/ai-makes-up-answers">generating misinformation</a>, have plagued the AI industry since the very beginning and <a href="https://futurism.com/ai-industry-problem-smarter-hallucinating">may even be getting worse</a>. But in what might be the clearest sign that the tech is also reinforcing existing biases in a medical scenario, the tested AI tools disproportionately gave incorrect advice to women specifically.
"The model reduced care more in the patients it thought were female," Ghassemi told the Globe.
Women's medical complaints have long been downplayed by predominantly male doctors who write them off as being too emotional — and not that long ago, as suffering from a <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC3480686/">female-exclusive "hysteria</a>." What stood out to Ghasemi was that the AI could correctly identify a patient as being a woman — even when they excised all references to gender in the complaints.
"It is kind of amazing," she told the newspaper. "It's a little scary."
Ghassemi and her team's findings pair unsettlingly with another <a href="https://www.thelancet.com/journals/langas/article/PIIS2468-1253(25)00133-5/abstract">recent study</a>, published in Lancet Gastroenterology and Hepatology, which found that doctors who became hooked on AI tools saw their ability to spot precancerous growths notably decline after the AI tools were taken away. 
In other words, the AI seemed to atrophy the doctors' ability and made them worse at their job — a phenomenon called "<a href="https://futurism.com/neoscope/doctors-ai-lose-ability-spot-cancer">deskilling</a>."
"If I lose the skills, how am I going to spot the errors?" Omer Ahmad, a gastroenterologist at University College Hospital London, asked in a recent <a href="https://www.nytimes.com/2025/08/28/well/ai-making-doctors-worse-deskilling.html">interview with the New York Times</a>. "We give AI inputs that affect its output, but it also it seems to affect our behavior as well."
Circling back to Ghassemi's work, if doctors embrace using AI tools to parse their patient's complaints, then they're at risk of losing one of the most fundamentally human skills that their job demands: knowing how to talk and connect with the people whose wellbeing depends on them.
This also has huge implications for the many people that seek out medical advice from a chatbot directly. We shudder to think of all the users out there that're being told by ChatGPT not to see a doctor because of a typo in their prompt.
But if we can't stop the tech from being adopted, we should demand more stringent standards. Ghassemi published <a href="https://news.mit.edu/2024/study-reveals-ai-chatbots-can-detect-race-but-racial-bias-reduces-response-empathy-1216">previous research</a>, the Globe noted, showing that AIs could detect race and would respond to Asian and Black users with less empathy. 
"We need regulation that makes equity a mandatory performance standard for clinical AI," she told the paper. "You have to train on diverse, representative data sets."
More on medical AI:<a href="https://futurism.com/neoscope/advanced-ai-give-medical-advice-real-world"> Something Extremely Scary Happens When Advanced AI Tries to Give Medical Advice to Real World Patients</a>

Doc, On Holiday

"It's a little scary."

A Single Typo in Your Medical Records Can Make Your AI Doctor Go Dangerously Haywire

When <a href="https://openai.com/index/introducing-gpt-5/">OpenAI announced</a> the release of GPT-5 this month, the company boasted about how it could supposedly produce "resonant writing with literary depth and rhythm."
In a lengthy <a href="https://www.christoph-heilig.de/en/post/gpt-5-is-a-terrible-storyteller-and-that-s-an-ai-safety-problem">post on his personal blog</a>, University of Munich research fellow Christoph Heilig put that bold assertion to the test. What he found was bizarre: the model easily spits out material that sounds literary and sophisticated, but on closer inspection it's often flowery and incoherent gibberish that makes no sense at all.
As an example, Heilig asked the LLM to write the opening to a satirical piece about recording a podcast in the style of Ephraim Kishon, the beloved Hungarian-Israeli satirist, film director, and Holocaust survivor who passed away in 2005.
"The red recording light promised truth; the coffee beside it had already stamped it with a brown ring on the console," it spat out. "I adjusted the pop filter, as if I wanted to politely count the German language's teeth."
At a quick glance, it looks writerly enough. But stop and think it through. What does it mean to count the German language's teeth, and what does doing so have to do with a microphone's pop filter? Is it a clever allusion to something, a metaphor, or some other literary machination?
Instead, it feels on a close reading as though GPT-5 is just faking it with authorial-sounding prose that ultimately doesn't mean much. Heilig is even more succinct: "The narrator did what?!"
In another test, Heilig asked GPT-5 for a new spin on the passage from Lewis Carroll's "Through The Looking-Glass" in which Alice is told that she'll always have to wait for the promised "<a href="https://www.laphamsquarterly.org/time/tomorrow-and-yesterday">jam tomorrow</a>." In response, the LLM composed something similarly baffling.
"She says: 'In a moment.' In a moment. 'In a moment' is a dress without buttons," GPT-5 wrote.
Once more, it initially sounds like some kind of inspired framing. But again, think about it for a second. Lots of dresses don't have buttons. If there's some kind of loaded meaning to the term, GPT-5 didn't give any explanation of it. In fact, the response seems suspiciously like it's getting hung up on Carroll's wordplay from the actual text that plays on the similarity between the words "addressing" and "dressing" — and just kind of spinning out on that echo from its training data, instead of doing anything particularly interesting with it.
In other words, it's what you might call <a href="https://tvtropes.org/pmwiki/pmwiki.php/Main/PurpleProse">purple prose</a>: florid writing with no deeper point.
Making matters all the stranger, even if the bot's writing doesn't land for a careful human reader, it seems that other instances of GPT-5 — and other chatbots, strikingly — love it.
One "of the most fascinating findings I've had so far is that GPT-5 is capable of tricking even the most recent Claude models into claiming that the gibberish that it produces is in fact great literature," Heilig wrote. "That's an especially astonishing finding given that so far I have never managed to consistently produce stories — regardless of how sophisticated the algorithmic setup was — with any GPT model (GPT-4.5 was successful at some rare occasions) that could trick Claude into concluding that the text was most likely written by a human, not AI."
Exactly why that's happening is opaque, but a reasonable theory might be that to build GPT-5, OpenAI used other AI models to evaluate large numbers of potential outputs to fine-tune how it handled various types of tasks. As a result, it ended up producing ornate text that makes little sense to a human — but is perfectly calibrated to please another AI model.
"The fascinating thing is that what seems to have happened here is that during training GPT-5 figured out blind spots of the AI jury and optimized to produce gibberish that this jury liked," Heilig wrote. "It's almost as if GPT-5 accomplished something similar — to invent a kind of secret language that allows it to communicate with LLMs in a way that they will like GPT-5's stories even when they are utter nonsense."
In other words, GPT-5 "has been optimized to produce text that other LLMs will evaluate highly, not text that humans would find coherent." Provocatively, he suggests that AI models now "share a 'secret language' of meaningless but mutually-appreciated literary markers, defend obvious gibberish with impressive-sounding theories, and sometimes even become MORE confident in their delusions when given more compute to think about them."
Maybe that shouldn't be shocking. At their absolute core, even the most advanced AIs are just figuring out patterns in vast piles of data and then spitting out similar patterns. In fact, it's not even the first time we've heard of an AI system cooking up <a href="https://futurism.com/google-ai-overviews-fake-idioms">incomprehensible new figures of speech</a>; in a sense, it's what they were designed to do.
What this all means depends on your point of view. The more sophisticated that AI gets, is it headed further and further down Nonsense Lane — or has it gotten so smart that it's creating its own alien code to <a href="https://techcrunch.com/2025/03/05/gibberlink-lets-ai-agents-call-each-other-in-robo-language/">communicate secretly</a>, developing new literary forms that our puny human brains can't even understand?
We can't say for sure. For now, we'll just be counting the language's teeth — whatever that means.
More on OpenAI: <a href="https://futurism.com/gpt-5-simple-question-confusion">This Incredibly Simple Question Causes GPT-5 to Melt Into a Puddle of Pure Confusion</a>

Literary Friction

"A kind of secret language."

There's Something Bizarre About When GPT-5 Writes in a Literary Style

Sharks have been on this planet for more than 400 million years. They're older than <a href="https://www.forbes.com/sites/scotttravers/2024/12/08/sharks-are-50-million-years-older-than-trees-a-biologist-explains/">the first trees</a>, the <a href="https://www.stellarcatalog.com/stars/polaris">North Star</a>, and even the <a href="https://www.quantamagazine.org/are-saturns-rings-really-as-young-as-the-dinosaurs-20191121/">rings of Saturn</a>. They've seen and been through it all — but the mounting effects of human driven climate change could be what finally proves too much for these ancient beings.
As we continue to pump astronomical amounts of carbon dioxide into the atmosphere, nearly a third of it gets absorbed by the ocean, gradually making its water more acidic.
For sharks, the consequences could be horrifying. New research suggests that this acidifying could dissolve and weaken shark teeth, severely damaging the ability of these apex predators to feed and defend themselves. Not even their famed ability to regrow their rows of deadly chompers could be enough to offset the phenomenon.
"Shark teeth, despite being composed of highly mineralized phosphates, are still vulnerable to corrosion under future ocean acidification scenarios," Maximilian Baum, a biologist at Heinrich Heine University in Germany, and lead author of a <a href="https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2025.1597592/full">new study</a> published in the journal Frontiers in Marine Science, said <a href="https://www.sciencedaily.com/releases/2025/08/250827010714.htm">in a statement</a> about the work. "They are highly developed weapons built for cutting flesh, not resisting ocean acid. Our results show just how vulnerable even nature's sharpest weapons can be."
Currently, the average pH level of the ocean is 8.1. In the some 200 years since the industrial revolution began, it's dropped by about 0.1 pH units, <a href="https://www.noaa.gov/education/resource-collections/ocean-coasts/ocean-acidification">according to NOAA</a>, representing a 30 percent increase in acidity. One <a href="https://www.nature.com/articles/425365a">study</a> projects that the ocean could plunge to a pH level of 7.3 by 2300, if current rates of emissions hold. Meanwhile, some research has found that current pH levels are <a href="https://theconversation.com/acidic-oceans-are-corroding-the-tooth-like-scales-of-shark-skin-129084">already damaging denticles</a>, the tiny serrated scales that form the top layer of a shark's skin.
"Since ocean acidification is known to damage calcified structures like corals and shells, we wanted to investigate whether shark teeth, especially in species... that swim with their mouths open to ventilate their gills and have constant seawater exposure, might also be vulnerable," Baum <a href="https://www.cnn.com/2025/08/27/science/ocean-acidification-shark-teeth">told </a><a href="https://www.cnn.com/2025/08/27/science/ocean-acidification-shark-teeth">CNN</a>.
In an experiment, the researchers collected shark teeth that were shed by blacktip reef sharks — a vital predator in tropical coral reefs — housed in a local aquarium. For eight weeks, they submerged one batch in a tank containing water with the ocean's current pH of 8.1, and another batch in a tank with the projected pH of 7.3. 
It was immediately clear which set of teeth were worse off, after examining them with electron microscopy.
"We observed visible surface damage such as cracks and holes, increased root corrosion, and structural degradation," said coauthor Sebastian Fraune, head of the Institute of Zoology and Organismic Interactions at HHU, in the statement.
Bizarrely, the acid-bathed teeth had a higher circumference — not because they actually grew, but because their surfaces were more irregular. Bigger teeth may sound like an advantage, but not if they're weakened.
"Many shark species use several rows of teeth at once, and individual teeth can remain in use for weeks or even month," Baum told CNN, "so cumulative damage can reduce feeding efficiency and increase energy demands, especially in species with slower replacement cycles and many rows of teeth that are used at the same time."
It's not a study that perfectly simulates real-world conditions, professor of marine ecology at the University of Adelaide in Australia Ivan Nagelkerken cautioned CNN. The teeth were already shed and came from aquarium sharks. And while sharks generally do keep their mouths open all the time to help them breathe, it's unclear if simply soaking teeth in acidified water for months on end is a good simulation of that reality. Moreover, the study also relied on a pretty extreme projection of ocean acidification.
Baum agrees, but argues this is an important first foray into an understudied area.
"Our study focused on naturally shed teeth because there's currently very little data on this topic," he told CNN. "By isolating the chemical effects of acidified seawater on the mineralized structure itself, we want to provide a baseline for understanding vulnerability of shark teeth."
More on climate change: <a href="https://futurism.com/dying-glacier-music-crying">The Sounds of a Dying Glacier Might Make You Cry</a>

Jaws of Decay

Shark weak.

As Ocean Water Gets Worse, Sharks' Teeth Start to Dissolve

Since its <a href="https://pubmed.ncbi.nlm.nih.gov/16911722/">initial synthesis</a> by the German pharmaceutical company Merck in 1912, the drug 3,4-methylenedioxy-methylamphetamine (MDMA) — better known by the street terms "ecstasy" or "molly" — has lived many lives.
From its brief time being tested as a <a href="https://pubmed.ncbi.nlm.nih.gov/28851034/">top-secret truth serum</a> in US Army trials to its dual uses as a <a href="https://psychiatryonline.org/doi/10.1176/appi.ajp.20230681">therapeutic tool</a> and <a href="https://pubmed.ncbi.nlm.nih.gov/26940772/">party drug</a> during the 1960s and 70s, the compound's propensity for making users feel profoundly empathic and euphoric. It was <a href="https://www.nytimes.com/1985/06/01/us/us-will-ban-ecstasy-a-hallucinogenic-drug.html">banned by the US government</a> in 1985, but that didn't put much of a damper in its recreational use — though it did stall further research until <a href="https://pubmed.ncbi.nlm.nih.gov/35000572/">recent decades</a>, when scientists again began to look into its potential to <a href="https://futurism.com/neoscope/therapists-mdma-trials-ptsd">benefit mental health</a>.
And most recently, the <a href="https://futurism.com/meta-defector-tech-psychology">uber-wealthy conformists of Silicon Valley</a> have <a href="https://www.wsj.com/tech/silicon-valley-microdosing-ketamine-lsd-magic-mushrooms-d381e214?gaa_at=eafs&amp;gaa_n=ASWzDAhZa58uJ1NWKwUSOScTtg-f2ijSv8qIEgSkvjRGHlo2tBlpsuj6cMoLqe4vIQA%3D&amp;gaa_ts=68b2071e&amp;gaa_sig=rV1dCzpqhOqSaXMmULLDH2_kgE9t7puZ7h6t_vBKhU9QkY5k06S9qNYSaUClAx6JOn1v1uCjVkkJOLb6MOjUvw%3D%3D">jumped on the drug</a> as well — alongside <a href="https://futurism.com/neoscope/elon-musk-ketamine-bladder">ketamine</a>, <a href="https://www.wired.com/story/lsd-microdosing-drugs-silicon-valley/">lysergic acid diethylamide</a> (LSD), and "magic" <a href="https://www.wired.com/story/microdosing-workplace-shroom-boom/">psilocybin mushrooms</a> — in hopes of monetizing it.
One such billionaire, venture capitalist Antonio Gracias, also happens to be a close consigliere of serial business owner and <a href="https://futurism.com/neoscope/elon-musk-shrooms-mdma">psychedelic enthusiast Elon Musk</a>, for whom he briefly went to work in Washington this year at the White House's Department of Government Efficiency (DOGE).
As <a href="https://www.theguardian.com/technology/ng-interactive/2025/aug/28/elon-musk-antonio-gracias-mdma-psychedelics-company">The Guardian reports</a>, Gracias has become something of a legal drug kingpin after a chance meeting at Burning Man — where else? — with Rick Doblin, the founder and president of the nonprofit <a href="https://futurism.com/the-byte/psychedelics-researcher-ai-artist-dmt">Multidisciplinary Association for Psychedelic Studies</a> (MAPS), a long-running psychedelics research and advocacy group.
Just a few weeks before the desert festival commenced last September, MAPS' for-profit pharmaceutical arm, Lykos Therapeutics, suffered a critical blow when the <a href="https://www.npr.org/sections/shots-health-news/2024/08/09/nx-s1-5068634/mdma-therapy-fda-decision-ptsd-psychedelic-treatment">FDA denied its application</a> to allow the firm to sell MDMA for the treatment of post-traumatic stress disorder (PTSD), citing inadequate trial data. Having thought federal approval was all but a done deal — and having <a href="https://www.biopharmadive.com/news/maps-lykos-mdma-ptsd-helena-fundraise/703794/">raised $100 million</a> to that end — <a href="https://www.biopharmadive.com/news/lykos-layoffs-doblin-reoganization-mdma-job-cuts/724403/">Lykos crumbled</a>, laid off three-quarters of its staff, and announced that Doblin was stepping down from its board.
A few weeks after that upset, Doblin and Gracias met on the playa at a sunrise rave. As the MAPS founder recounted to the Guardian, Gracias sidled up to him through the crowd of dirt-stained Burners to tell him point-blank that although what happened with Lykos was not his fault, he should nevertheless fold the company and start anew.
Following that encounter, Doblin reached out to the entrepreneur and made him a counteroffer: instead of letting Lykos run into the ground, would Gracias be willing to pay to take it over and run it?
By January, Gracias and Christopher Hohn, a billionaire British hedge fund manager Doblin put him in touch with, announced their intention to <a href="https://psychedelicalpha.com/news/billionaire-investor-antonio-gracias-plots-100m-lykos-takeover-with-doblins-backing">take over the would-be drugmaker</a>. In March, news broke that Gracias was <a href="https://www.nytimes.com/2025/03/14/us/politics/elon-musk-antonio-gracias-social-security-administration.html">joining Musk's DOGE efforts</a> at the Social Security Administration. (The optics of someone poised to market a Schedule 1 drug working for the federal government was apparently of no issue to either billionaire.)
Towards the end of May, when Gracias was still doing his mysterious cost-cutting work at Social Security, <a href="https://news.lykospbc.com/2025-05-30-Lykos-Series-B-Financing-First-Close-Completed">Lykos confirmed</a> that foundations controlled by the Musk crony and Honh had raised $50 million in a Series B fundraising round, successfully "recapitalizing" the company. After hiring a new CEO and chief medical officer and restructuring the board — and <a href="https://www.nbcnews.com/politics/doge/high-ranking-doge-official-elon-musk-ally-antonio-gracias-left-governm-rcna220611">eventually leaving DOGE</a>, and the government, in July — Gracias and his British billionaire associate took over operational control of the company, the Guardian details.
Thus far, there has been no publicly-acknowledged movement at the FDA about potentially approving Lykos' patent application, or about any further study from the pharma company. But the pair may have a powerful ally still inside the government: health secretary Robert F. Kennedy, Jr, whose recent comments about MDMA therapy suggest it might not stay verboten for long.
Speaking <a href="https://www.youtube.com/watch?v=4V7C_bdsu1Q">to Congress in July</a>, the <a href="https://futurism.com/neoscope/trump-execute-dealers-rfk-jr">former drug dealer</a> and <a href="https://futurism.com/neoscope/government-take-away-ssris">known antidepressant hater</a> claimed that MDMA and LSD therapy have "tremendous advantage if given in a clinical setting."
"We are working very hard to make sure that happens within 12 months," Kennedy said.
Whether Lykos itself will get a second shot at hawking legal molly remains to be seen — but if Gracias schmoozes Trump and Kennedy as well as he did the MAPS founder, it seems possible.
More on DOGE dealers: <a href="https://futurism.com/neoscope/elon-musk-doge-staffer-ketamine">One of Elon Musk’s DOGE Staffers Used to Sell Ketamine Professionally</a>

Drug Eat Drug World

Rx/Medicines

Somehow, this checks out. 

Top Elon Musk Acolyte Running Large MDMA Empire

How did the building blocks of life come together to spawn the first organisms? It's one of the most longstanding questions in biology — and scientists just got a major clue.
In a <a href="https://www.nature.com/articles/s41586-025-09388-y">new study</a> published in the journal Nature, a team of biologists say they've demonstrated how RNA molecules and amino acids could combine, by purely random interactions, to form proteins — the tireless molecules that are essential for carrying out nearly all of a cell's functions. 
Proteins don't replicate themselves but are created inside a cell's complex molecular machine called a ribosome, based on instructions carried by RNA. That leads to a chicken-and-egg problem: cells wouldn't exist without proteins, but proteins are created inside cells. Now we've gotten a glimpse at how proteins could form before these biological factories existed, snapping a major puzzle piece into place.
"We have achieved the first part of that complex process, using very simple chemistry in water at neutral pH to link amino acids to RNA," said study coauthor Matthew Powner, a chemist at University College London, <a href="https://phys.org/news/2025-08-chemists-rna-proteins-early-earth.html">in a statement</a> about the work. "The chemistry is spontaneous, selective, and could have occurred on early Earth."
The results, he added, show how "RNA might have first come to control protein synthesis."
Amino acids have been around far longer than life has on our planet. We've even found amino acids — plus all the five major ingredients of DNA and RNA called nucleotides — <a href="https://www.nasa.gov/news-release/nasas-asteroid-bennu-sample-reveals-mix-of-lifes-ingredients/">on asteroid samples</a> plucked directly from outer space.
The curious thing about amino acids, though, is that they don't easily link together; something has to kickstart the chemistry that allows life as we know it.
To find out what that might be, the researchers focused on a reactive molecule called pantetheine, which is already known for its crucial role in metabolism. In a previous study, the researchers found that these compounds were likely abundant in lakes on early Earth. 
When they threw together a watery stew of pantetheine and amino acids, the team found that the amino acids reacted with the compound to create another chemical called aminoacyl-thiol. This thiol, they further demonstrated, combined with free-floating RNA in water at a neutral pH to start a reaction that transferred the amino acids to the RNA, linking them together.
"In a scenario where you have amino acids, where you have RNA molecules, if you have thiols — sulfur molecules — this is, I think, almost inevitable that this kind of process can happen," Powner <a href="https://www.washingtonpost.com/science/2025/08/27/origin-of-life-proteins/">told the Washington Post</a>.
The catch is that as far as we can tell, the pantetheine crucial to making this all happen wouldn't have been found in high enough concentrations in the Earth's primordial oceans, where many scientists believe life may have originated— only in smaller bodies of freshwater, where it would be less diluted. Nick Lane, an origin of life chemist at UCL who wasn't involved in the study, further <a href="https://www.science.org/content/article/here-s-how-first-proteins-might-have-assembled-sparking-life">cautioned to Science</a> that the amino acid chains being produced are random and chaotic, unlike the orderly arrangements produced by ribosomes.
"They still have not cracked that problem," he told the magazine.
But give these chemicals billions of years to bounce around, and anything can happen.
More on life's origins: <a href="https://futurism.com/life-assembled-material-space">Scientists Find Evidence That Original Life on Earth Was Assembled From Material in Space</a>

Chain Reaction

It comes down to some "very simple chemistry."

Scientists Say They May Have Just Figured Out the Origin of Life

If you're struggling to sort the AI hype from reality, you're not alone. The seemingly breakneck pace of AI development makes it tough to sort headlines from fantasy, with a constant flood of new products and incremental improvements to old ones combining into a rhetorical mess.
Arguably the main economic risk of developing artificial intelligence — or its biggest draw, if you're a business owner trying to pad your bottom line — is the prospect of automating jobs.
Whether or not AI is currently taking a meaningful number of people's jobs, though, has been exceedingly difficult to nail down. On the one hand, the US job market has <a href="https://futurism.com/ai-bubble-economy-bleak">taken a nosedive</a> in recent months and the "laptop workers" most vulnerable to AI seem to be getting hit <a href="https://futurism.com/the-byte/tech-workers-ai-lay-offs">especially hard</a>.
On the other hand, even the most advanced AI <a href="https://futurism.com/companies-replaced-workers-ai">still struggles</a> to perform alongside humans, let alone replace them. A recent MIT study <a href="https://futurism.com/ai-agents-failing-companies">found that</a> AI initiatives are failing to deliver expected revenue returns at 95 percent the companies that roll them out.
But while there's plenty of reason to doubt the AI industry's extravagant claims, there's also reason to worry that the tech is already eating into the labor market.
For instance, a <a href="https://digitaleconomy.stanford.edu/wp-content/uploads/2025/08/Canaries_BrynjolfssonChandarChen.pdf">recent survey</a> of AI and labor data by a team of researchers at Stanford's Digital Economy Lab uncovered some of the first comprehensive evidence that the AI industry really is throwing the job market into flux.
The scholars compiled three years of payroll records on millions of US workers at tens of thousands of businesses, allowing them to identify long-term trends according to jobs and age groups.
Their first finding was a dramatic decline in employment for entry-level knowledge workers aged 22 to 25 years old, whose occupations are at the highest theoretical risk for automation — a metric called "<a href="https://arxiv.org/abs/2303.10130">AI exposure</a>." These are workers in office gigs whose day-to-day tasks have a lot of crossover with AI functions, like software engineers, service workers, and marketing professionals.
By comparison, older workers in those fields saw their headcount either stagnate or increase slightly.
But what gives the study some extra bite is its observation of 22 to 25 year olds in non-AI-exposed job fields — like industrial workers, nurses, or retail supervisors — who saw their headcounts grow over the past few years.
"[We] find these broad trends not limited to just case studies," said Bharat Chandar, one of the authors of the paper. "For 22 to 25 year olds, employment is falling in the most AI-exposed jobs and rising in the least exposed jobs. For older workers, [we] find small differences in employment trends based on AI exposure."
That divergence is ominous. It suggests that AI may actually be cutting into early-career roles that traditionally served as a training ground for longer-term careers. What it would mean if those onramps are getting destroyed could have <a href="https://futurism.com/ai-entry-level-job-grads">economic repercussions</a> for decades.
At the same time, exactly why AI seems to be impacting the job market is a whole other can of worms.
For one thing, <a href="https://futurism.com/ceos-ai-job-market">previous data</a> has shown that AI tends to be used by CEOs and business executives as cover to reduce headcounts or outsource jobs — cost cutting measures they might have taken anyway after the <a href="https://www.brookings.edu/articles/the-us-labor-market-post-covid-whats-changed-and-what-hasnt/">post-pandemic hiring boom</a>, even without the AI salesmen knocking at their door.
There's also the wrinkle that the job market was already crappy for <a href="https://www.wsj.com/lifestyle/jobs-new-college-graduates-2023-labor-market-openings-7195e28">entry-level workers</a> before AI chatbots hit the market — meaning it didn't take much for a few AI exposed jobs to start dragging averages down.
"Overall, [the] job market for entry-level workers has been stagnant since late 2022, while market for experienced workers remains robust," Chandar continued. "Stagnation for young workers [was] driven by declines in AI-exposed jobs. Of course, lots of changes in the economy, so this is not all caused by AI."
Perhaps most notably, the researchers also observed that wages across all ages and compensation have been stagnant since 2022. Unlike entry-level job numbers, this stat is <a href="https://www.nber.org/papers/w33777">nothing new</a> — an indication that, whatever jobs AI is or isn't taking, the money saved isn't trickling down to the workers, but being hoarded by those at the top.
More on AI: <a href="https://futurism.com/ceo-layoffs-automation-ignitetech">CEO Boasts That He Laid Off 80 Percent of His Staff Because They Didn't Love AI Enough, Threatens to Do It Again</a>