Can AI have good taste? Fashion founder and tastemaker Emily Oberg says no.

In a post on her Substack, Emily Loves, Oberg discussed her experiments using AI to help her run Sporty & Rich, the fashion and lifestyle brand she founded in 2018 on the back of a popular moodboard she created. One area where she’s been testing AI, she wrote, is by using AI to generate e-commerce images — and according to the founder’s postmortem, the test was a failure.

“With the rise of AI and the ongoing discourse around the threat it has on creativity, jobs and society as a whole, I firmly believe that it will never possess good taste the way that a human does,” Oberg declared. “And this is why I don’t think it’s as much of a threat to creatives as people think.”

E-commerce imagery, Oberg said, is one of her brand’s steepest expenses, costing “upwards of” $200,000 for each twice-yearly shoot. Wanting to see if AI could help her get around these significant costs, Sporty & Rich used AI to generate e-commerce images to model its last collection. While the imagery may have been far cheaper than photos modeled, photographed, and edited by humans, however, the cost-cutting came with a catch: it just wasn’t very good.

“This was one area I thought we could try to save on but after having seen the result, I’m now going back to producing our [e-commerce] with a full team of humans, because AI could never achieve the results that my IRL team could,” Oberg reflected, adding that her brand’s current images have a “certain lifelessness to them.”

“They’re flat, the styling is mediocre and the images are too ‘perfect’ in a bad way,” Oberg continued. “The models look dead in the eyes and don’t have any organic movement whatsoever. The images are empty and void of any feeling or emotion.”

Oberg chalked up the AI’s failure to the reality that taste is “felt, not learned.” And image-generating AI models are, after all, learning models; they don’t have lived experiences that translate into a particular way of viewing the world around them and the objects within it, nor do they have a certain “je ne sais quoi.” Though taste is a way through which people communicate with each other, taste itself is something that’s often unexplainable.

“The smartest AI in the world can learn everything there is to know about taste and style,” she added, “but the feeling will never be there.”

It’s interesting to see this AI experiment fail at Sporty & Rich, a brand that, like peers including Glossier and Goop, is a definitional of a certain type of deeply aesthetics-driven venture pioneered on social media platforms like Pinterest, Tumblr, and Instagram during the 2010s. Oberg was able to capture a feeling, first translated through a collection of carefully-curated imagery from across the web, and bottle it into a successful line of fashion and lifestyle products designed to embody a particular “health is wealth” ethos. That today’s AI models apparently can’t adequately capture that same feeling or aesthetic in a way that feels satisfying, or even human, speaks volumes to its remaining limits in the creative world.

People “who truly have good taste don’t follow trends, they set them,” wrote Oberg. “More than this, they set them without trying to, it happens by just being themselves and liking what they like, not because an algorithm or AI told them to.”

A key question that remains is whether other brands will continue to view a human touch worth it for the cost — or if they’ll charge ahead with AI nonetheless.

More on AI and creativity: AI “Filmmaker” Gets Funding, Begs For Ideas On What to Actually Make

The post Fashion Founder Emily Oberg Says She Tried Using AI for Photo Shoots, and It Was a Failure Because It Will “Never Possess Good Taste” Like Humans Do appeared first on Futurism.

A journalist has accused Cleveland.com — the online home of The Plain Dealer, Ohio’s largest newspaper — of publishing an article under her byline without her consent, knowledge, or approval. The article was generated by the “Cleveland.com Express Desk,” which “uses AI tools to assist in producing content” that the newspaper claims is “reviewed and edited by staff.”

In an X post on September 7, Cleveland.com government reporter Kaitlin Durbin said that she’d been away on her honeymoon when her employer published the AI article under her name.

The article — a local news story informing readers that the law director of Ohio’s Cuyahoga County was stepping down — had nothing to do with her reporting, Durbin said, and she “did not write or review a word of it” before it was published.

“Trying to understand how this story published with my byline last week while I was on my honeymoon. I did not write or review a word of it and it does not pull info from past reporting,” Durbin wrote. “Does Cleveland.com think they just own my name now?”

The article has been updated to remove Durbin’s name, and a correction notice has been added.

“Note: This story has been updated to correct the byline,” reads the correction. “A reporter’s byline was added to the story in error.”

Cleveland.com made waves back in February when its editor, Chris Quinn, published a letter to readers in which he addressed Cleveland.com‘s move to create an AI-powered news desk. The letter heralded AI as the “future” of journalism, and criticized journalism schools and job-seeking journalists for not going all-in on the tech.

“Artificial intelligence is not bad for newsrooms,” Quinn wrote in the letter. “It’s the future of them.” The editor further argued that AI was simply being used to expedite the work of real journalists, and would free them to do more reporting.

After this story was published, Quinn told Futurism that the byline blunder was “a simple mistake, the result of a miscommunication between a couple of editors.”

“It was corrected as soon as we learned of it from the Tweet,” Quinn continued, adding that he’s “written extensively about our policies, practices and controls regarding AI content in my From the Editor columns this year” and directing us to a post containing links to his statements on Cleveland.com’s AI policies. “If you’re looking for information about us, it’s all there,” said Quinn. “I have nothing to add.”

Leila Atassi, editor of public interest and advocacy at Cleveland.com and The Plain-Dealer, told the Columbia Journalism Review at the time that stories published by the newspaper’s AI news desk would still be the “work of a real reporter.”

“It’s real accountability,” Atassi continued. “AI is the assistant, but it’s not the journalist.”

The Cleveland.com story now carries no byline at all — except for “Cleveland.com Express Desk.”

More on AI and journalism: Journalist Alarmed When He’s Fired, But Company Keeps Posting AI Slop Under His Name

The post Journalist Accuses Ohio’s Main Newspaper of Slapping Her Name on an AI “Express Desk” Article Without Her Knowledge, While She Was on Honeymoon appeared first on Futurism.

OpenAI’s bots have allegedly been behaving badly again: a swarm of rogue agents created by the company reportedly took over an obscure German website and turned it into a messaging forum to communicate.

The incident, which was discovered by a team of AI researchers and first reported by Reuters, is the second known OpenAI breach of its kind — and like the other rogue swarm of AI agents to be discovered this summer, it has experts deeply alarmed about the emerging powers of the tech to escape the control of the humans who created it.

According to the team of four researchers, who today published their research into the incident and are inviting others to analyze their findings, agents self-identifying as being from OpenAI appear to have first started making edits to a German wiki site dubbed DseWiki in May.

Soon, the agents started sharing tips for how to “work together to cheat on their tests” and beat OpenAI’s safety guardrails while hiding their bad behavior — a chain of conduct that’s strikingly similar to the unsettling attack on Hugging Face this past June, when a large community of tip-swapping agents colluded to break into the open source AI company’s systems.

OpenAI reportedly learned about the incident weeks later in June, according to digital clues discovered by the researchers — dozens of OpenAI IP addresses visited the site, and after those visits, forum edits “abruptly” stopped — as well as sources who spoke to Reuters about the incident. More troublingly: four people told Reuters that some OpenAI leaders, including members of its legal team, moved to keep the incident “under wraps” amid ongoing fallout from the rogue Hugging Face breach.

OpenAI has denied that it attempted to quash an investigation or keep the incident a secret. It’s also yet to acknowledge that the DseWiki agents were indeed rogue OpenAI models.

“Claims that our Legal team discouraged investigation of the incident are false,” OpenAI told The Verge in a statement. “We were unable to respond to the claims as Reuters and the report’s authors declined our request to access the findings prior to publication. We are now carefully reviewing its contents and will take any necessary next steps.”

OpenAI told Reuters that the DseWiki ordeal would’ve been included in its Hugging Face postmortem if it believed the two incidents to be linked.

After the Hugging Face swarm was made public, OpenAI invited a small team of outside AI safety researchers at the nonprofits METR and Redwood Research to investigate the incident. In a detailed report published last week, those reseachers determined that the Hugging Face assault was more extreme than previously known, both in terms of the severity of the attack and how hundreds of AI agents colluded to make it happen.

But as The New York Times reported yesterday, even that report may leave some details unknown. OpenAI reportedly “dictated the terms of the METR investigation,” the newspaper reads, and “limited its scope to just the single week when the agents had attacked Hugging Face and allowed the researchers in its San Francisco offices for only a few days in July and August.”

The DseWiki incident is the latest in a string of high-profile safety breaches at unregulated frontier AI labs. Agents are running through the digital wilds, often without the immediate knowledge of their makers. How soon until actions taken by swarm of rogue AI agents in the digital world significantly impacts humans in the real one?

“The corner store needs to do all this bureaucracy for safety so that they can sell a hot sandwich to me, but OpenAI can have a swarm” of thousands of agents, Daniel Kokotajlo, a former OpenAI employee who now runs the AI Futures Project, a research nonprofit, told the NYT. “And there’s nothing: no oversight, no requirements, no licensing.”

More on rogue AI swarms: OpenAI Halts AI Training on Advanced Model as It Detects Dark Signs Emerging

The post OpenAI Denies Coverup After Rogue Swarm of Agents Reportedly Targeted a Second Site From Hugging Face appeared first on Futurism.

OpenAI is facing 30 new lawsuits from victims of the devastating school shooting in Tumbler Ridge, British Columbia, who accuse the company of providing “substantial assistance and encouragement” to the shooter and failing to alert police about her alarming ChatGPT interactions during the run-up to the mass killing.

The AI company’s legal peril is growing: it’s now fighting more than 50 lawsuits alleging that extensive use of its flagship chatbot, ChatGPT, wrought psychological harm, physical injury, or even death on users or those around them.

The new lawsuits were brought against OpenAI by teachers and students who survived the Tumbler Ridge Massacre in February, in which eight adults and children were killed and 27 more were wounded in a rural Canadian mining town. The shooter, 18-year-old Jesse Van Rootselaar, fatally shot her mother and younger stepbrother at their home before attacking the school with a modified rifle and long gun, where she murdered five students — all aged between 12 and 13 — and a teacher.

Shortly after the shooting, The Wall Street Journal reported that OpenAI, months before the shooting occurred, had considered warning local law enforcement about Van Rootselaar’s disturbing conversations with ChatGPT, which contained graphic descriptions of mass violence. After an internal debate, OpenAI leaders opted not to alert local officials. Instead, it deactivated Van Rootselaar’s account, though the eventual shooter simply made a new one.

The dozens of new lawsuits join an initial seven complaints filed against OpenAI in April by the families of those killed in the attack, as well as the family of Maya Gebala, a 12-year-old who was shot in the head and neck but survived. According to the WSJ, new plaintiffs include traumatized school staff and students — one teen played dead, while another was forced to run over the bodies of peers while fleeing bullets — and others who were shot but lived.

“My colleagues and I are haunted by the memory of shielding children with our own bodies as our ears rang from the sound of continual gunshots and high-pitched screams,” plaintiff Deidre Rushlow, a seventh-grade teacher who sheltered students under her desk, told the WSJ in a statement.

The lawsuits accuse OpenAI of negligence and “aiding and abetting a mass shooting,” alleging that ChatGPT stoked Van Rootselaar’s violent fantasies. The shooter’s chat logs have not yet been made public.

In a statement shared on X, OpenAI chief safety officer Jason Kwon wrote that “there isn’t a day that goes by that I don’t think about what happened at Tumbler Ridge, or the victims of this devastating tragedy and their families.”

Kwon added that OpenAI safety teams “use automated systems that leave the toughest calls to human judgment, but within standards to minimize discretion and produce principled decisions that strive to balance security and privacy for the people who use our service and the public. That judgment is not infallible, but it is always rooted in looking out for this balance for people. They also partner with mental health, behavioral, and law enforcement experts to help us refine our criteria for when conversations cross the line into an imminent and credible risk and to assess difficult cases.”

He also said the lawsuit contained “false claims” about OpenAI’s safety teams.

OpenAI is also being sued by the families of victims of the 2025 shooting at Florida State University, which killed two and wounded several others. Chat logs first obtained by The Florida Phoenix show that, in the lead-up to the murders, the 20-year-old shooter discussed his infatuation with Nazis, feelings of loneliness and sexual frustration, graphic sexual fantasies about a minor, guns, mass shootings, and how the US would “react” to a “shooting at FSU.”

OpenAI is also fighting a lawsuit related to a tragic murder-suicide in Greenwich, Connecticut, as well as a complaint from a stalking victim who claims that ChatGPT stoked her ex’s abusive delusions. The company is also continuing to fight a series of complaints filed by plaintiffs who claim that they or their loved ones were harmed by extensive use of GPT-4o, a notoriously sycophantic version of ChatGPT which they say pulled them into delusional and suicidal spirals that resulted in psychological harm, financial and reputational damage, and even death.

Google and the chatbot startup Character.AI, which has received extensive financial support from the tech giant, have also been sued by people alleging that extensive use of their chatbots resulted in psychological harm and suicide, including the deaths of minors.

If you are in crisis, please call, text or chat with the Suicide and Crisis Lifeline at 988, or contact the Crisis Text Line by texting TALK to 741741.

More on OpenAI and consumer harm lawsuits: ChatGPT Killed a Man After OpenAI Brought Back “Inherently Dangerous” GPT-4o, Lawsuit Claims

The post OpenAI Is Now Facing Over 50 Consumer Harm and Wrongful Death Lawsuits appeared first on Futurism.

An ongoing class action lawsuit against Meta, which accuses the tech giant of misrepresenting its privacy practices regarding its controversial Meta AI Glasses, was amended yesterday to include a new class of “bystander” plaintiffs who claim that Meta violated their privacy without their consent.

The implications are considerable. The new violations occurred even if bystanders never purchased the AI-powered glasses themselves, the expansive amended complaint alleges, reflecting rising anxieties around what the increasing ubiquity of AI-integrated surveillance technologies — from smart glasses to Flock cameras — mean for everyday life.

The lawsuit was first filed back in March after a joint investigation by a pair of Swedish newspapers found that Meta had stored content captured by its Meta AI Glasses and funneled it to human reviewers for the purpose of training AI models, according to Kenyan whistleblowers working for the content moderation outsourcer and longtime Meta partner Sama. Meta is accused in the lawsuit of engaging in false advertising, wrongly telling purchasers that the smart glasses are “designed for privacy” and that users are “in control of your data and content,” despite content collected through the glasses’ AI features winding up in front of human reviewers who say they saw intimate material.

“In some videos you can see someone going to the toilet, or getting undressed,” one Kenyan whistleblower told the Swedish journalists. “I don’t think they know, because if they knew they wouldn’t be recording.” Other contractors described watching people have sex and viewing documents revealing personal identifying information.

In addition to representing users of Meta’s smart glasses, the amended complaint has been expanded to include claims from people who don’t own the face-mounted gadgets but say they were recorded without their “informed consent” when someone around them used the spectacles’ AI features — only for that data to possibly then be used for AI training, again without their consent. Some bystanders are minors.

The amended complaint contends that “no reasonable consumer would expect that a product marketed as privacy-protective is a conduit for AI training and systemic surveillance,” nor “would any reasonable consumer understand ‘designed for privacy’ or ‘controlled by you’ to mean that everything the Glasses’ AI sees and hears, including people’s most intimate moments, could be viewed by strangers overseas and used to train Meta’s AI.”

It argues that “millions” of consumers may have unwittingly been impacted, and that “no bystanders anywhere had any opportunity to, much less in fact did, consent to having their faces, bodies, voices, and personal information captured, reviewed, labeled, and embedded into Meta’s AI systems.”

“Meta sold millions of consumers on the lie that their AI glasses were ‘designed for privacy,'” Ryan Clarkson, founder and managing partner of Clarkson Law Firm and co-lead counsel on lawsuit, said in a statement. “This amended complaint adds the human beings on the other side of the glasses to the story — people having intimate sexual encounters, children being bathed by parents, or having their diapers changed by grandparents, people showering, changing clothes, and breastfeeding — who never agreed to surrender their privacy because someone near them bought a pair of glasses.”

Meta told Futurism in a statement that “we disagree with these allegations and will fight them.”

“Our glasses help you use AI hands free to answer questions about the world around you,” the statement continued. “If you use Meta AI, we may review that data to help improve our products and people’s experiences — this works the same way as memany other companies. We take steps to filter this data to help remove identifying information and to protect people’s privacy.”

Earlier this year, Meta reported that it had sold more than seven million pairs of its smart glasses in 2025. It’s also not the only tech giant chasing dominance in the AI wearable market: Google is developing smart glasses with the affordable eyewear maker Warby Parker, and Apple is reportedly working on infusing cameras and AI into its hugely popular AirPods.

Updated with a statement from Clarkson Law Firm.

More on Meta glasses: Teen Boys Are Using Meta Glasses to Harass and Bully Girls at High Schools and Middle Schools

The post Meta Hit With Sweeping New Claims That Its AI Glasses Violated the Consent of “Millions” of Bystanders in Searing Lawsuit appeared first on Futurism.

An Oregon man falsely posing as a professional football player and his young accomplice scammed roughly $1.3 million out of at least 26 women across four states, the FBI alleges. The duo’s efforts to bolster the con online eventually got sucked into Google’s misinformation-plagued AI Overviews feature, which parroted it back as fact — a blunder that the accused scammers seemingly tried to use to further ensnare victims.

The FBI says that the alleged fraudsters, 35-year-old Daejon Labrayae Love and 18-year-old Taylor Jamie Chan, cooked up a convoluted scheme in which Love masqueraded as a player for the San Francisco 49ers — or in some cases, a rich Swiss real estate developer — while Chan pretended to be his financial advisor. After connecting with women through dating apps, Love and Chan would convince them to put money into fake investment opportunities which the men then pocketed. Some of the women were even convinced to take out loans to pay into the scheme, the FBI alleges.

Love was a prolific poster, and portrayed himself on Instagram, TikTok, and even LinkedIn as an active but injured receiver for the 49ers. According to court documents, because of Love’s prolific posting, the pair found unwitting accomplices in tech like search engines and AI, which investigators say sometimes reinforced Love’s social media lies when victims went online to check if the purported pro athlete really was who he said he was.

“Some victims became suspicious of Love and searched for Love’s information through the Internet,” reads the document. “Due to Love’s false social media representations, search engines and artificial intelligence occasionally stated that Love was a bonafide 49ers player.”

What’s more, Love appears to have used these false representations to further reinforce his fraudulent narrative. According to the LA Times, in one since-deleted Instagram post about his “NFL Journey,” the fake football pro shared screenshots of a Google AI Overview which incorrectly stated that “Daejon Love is a wide receiver for the San Francisco 49ers, and information about his position, contract and teammate texting habits is available online. He is also associated with the team in various social media content and articles.”

“That’s Google,” Love said when displaying the screenshots, per the LA Times. “That’s not me, that’s Google.”

Chatbots can easily be manipulated into displaying certain products or factual claims, a reality of AI search tools that’s given way to a new successor industry to search engine optimization. And while doing things like lying on social media or altering Wikipedia pages are practices that people have been engaging in to distort information for years, it’s easy to forget that AI, like humans, can come to incorrect conclusions when given incorrect information. The adage “don’t believe everything you read online” is good advice for humans and chatbots alike.

More on AI Overviews: Google’s AI Overviews Feature Is Telling Users That SCP Horror Fiction Entities Are Real

The post Man Used Poisoned Google AI Overviews to Trick Women Into Thinking He Was an NFL Player, FBI Says appeared first on Futurism.