The Sun — usually so predictable — is exhibiting some surprising behavior and that has scientists very intrigued.
Astronomers had predicted that our host star was entering a period of relative quiet back in 2008, but NASA scientists have published a <a href="https://iopscience.iop.org/article/10.3847/2041-8213/adf3a6">new study</a> in The Astrophysical Journal Letters that found that the Sun has instead defied expectations by becoming more active, with increased sunspots and solar flares.
"All signs were pointing to the Sun going into a prolonged phase of low activity," said the study lead author and NASA physicist Jamie Jasinski in a <a href="https://www.nasa.gov/science-research/heliophysics/nasa-analysis-shows-suns-activity-ramping-up/">statement about the paper</a>. "So it was a surprise to see that trend reversed. The Sun is slowly waking up."
Back in the 1980s, scientists had noticed the Sun's surface was steadily quieting down until approximately the 2008 financial crash.
That was the year that the celestial ball of plasma notched its "weakest on record" solar activity — which prompted scientists to predict that this relative state of quietude would keep on extending.
But like a tame camp fire that suddenly crackles into flames and sparks, the Sun abruptly reversed course during the Obama administration with "various plasma and magnetic field parameters" jumping up and the number of sunspots outstripping predictions.
The Sun has long been known to go through periodic cycles of activity every 11 years, <a href="https://www.esa.int/Science_Exploration/Space_Science/The_solar_cycle_a_heartbeat_of_stellar_energy">called a solar cycle</a>, when it becomes more active — eventually reaching a "solar maximum," in astronomy parlance — before powering down to a "solar minimum."
The 11-year period is part of a longer cycle, dubbed the Extended Solar Cycle or <a href="https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2022.923049/full">Hale cycle</a>, that averages around 22 years, and in which the Sun's magnetic poles reverse.
But the Sun also goes through longer spans of quiet, such as from 1645 to 1715, and 1790 to 1830, when sunspots basically disappeared. Given the Sun's long quieting period since the 1980s, scientists had anticipated that was due to happen again — but instead, it seems to be flaring back up.
"We don’t really know why the Sun went through a 40-year minimum starting in 1790,"said Jasinski. "The longer-term trends are a lot less predictable and are something we don’t completely understand yet."
Why should anybody on Earth care? Increased <a href="https://science.nasa.gov/sun/solar-storms-and-flares/">solar activity impacts our planet</a> by messing up radio communications, power grids, and navigation systems.
And with the United States and China competing in exploring space, monitoring solar activity is important to make sure <a href="https://www.skyatnightmagazine.com/space-missions/solar-storms-mars-maven-curiosity">space travel is safe from any surprising solar conditions</a> that can damage spacecraft and harm astronauts.
Plus, it's just cool. Ultimately, the Sun is what keeps all life on our planet going, instead of being a frozen husk in space — so its riddles are a matter of existential curiosity.
More on the Sun: <a href="https://futurism.com/secretive-plan-blocking-sun-geoengineering">Scientists Secretly Working on Plan to Test Blocking Sun From Huge Area of Earth</a>

Sunny Side Up

"The Sun is slowly waking up."

Something Weird Is Going on With the Sun, Scientists Find

Parents who allege their children were abused, physically harmed, and even killed by AI chatbots gave emotional testimonies on Capitol Hill on Tuesday <a href="https://www.judiciary.senate.gov/committee-activity/hearings/examining-the-harm-of-ai-chatbots">during a hearing</a> about risks to young users posed by the tech — all while urging lawmakers to enforce regulation in a landscape that remains a digital Wild West.
There were visible tears in the room as grieving parents <a href="https://futurism.com/parents-testifying-us-senate-ai-children">recounted their painful stories</a>. According to the lawmakers on the US Senate Judiciary Subcommittee on Crime and Terrorism, the bipartisan committee that held the session, representatives from AI companies declined to appear. The bipartisan panel laid into them in absentia — with the overwhelming consensus between lawmakers and testifying parents being that the AI industry has prioritized profits and speed to market over the safety of users, particularly minors.
"The goal was never safety. It was to win a race for profit," said Megan Garcia, whose son, Sewell Setzer III, died by suicide after extensive interactions with chatbots hosted by the Google-backed chatbot company Character.AI. "The sacrifice in that race for profit has been, and will continue to be, our children."
Garcia was joined by a Texas mother identified only as Jane Doe, who alleged that her teenage son suffered a mental breakdown and began to self-mutilate following his use of Character.AI. Both families have <a href="https://www.nytimes.com/2024/10/23/technology/characterai-lawsuit-teen-suicide.html">sued</a> <a href="https://futurism.com/character-ai-google-test-ai-chatbots-kids">Character.AI</a> — as well as its cofounders Noam Shazeer and Daniel de Freitas, and Google — alleging that Character.AI chatbots sexually groomed and manipulated their children, causing <a href="https://futurism.com/google-character-ai-children-lawsuit">severe mental and emotional harm</a> and, in Setzer's case, death. (In response to litigation, Character.AI built in <a href="https://futurism.com/character-ai-parental-controls-bypass">reactive parental controls</a>, and has <a href="https://futurism.com/character-ai-plans-teens">repeatedly promised</a> <a href="https://futurism.com/character-ai-emailing-minors-school-shooters">strengthened guardrails</a>.)
At the time that both teens downloaded the app, it was <a href="https://futurism.com/character-ai-google-play-best">rated safe for teens</a> on both the Apple and iOS app stores. Though it's declined to publicly share information about safety testing, Character.AI continues to market its product as safe for teens. There's currently no regulation preventing the company from doing so, or compelling chatbot makers to make information about their guardrails and safety testing public. On the morning of the hearing, <a href="https://www.washingtonpost.com/technology/2025/09/16/character-ai-suicide-lawsuit-new-juliana/">The Washington Post reported that</a> yet another wrongful death suit, this one for a 13-year-old girl who died by suicide, had been filed against Character.AI.
"I have spoken with parents across the country who have discovered their children have been groomed, manipulated, and harmed by AI chatbots," said Garcia, warning that her son's death is "not a rare or isolated case."
"It is happening right now to children in every state," she added. "Congress has acted before when industries placed profits over safety, whether in tobacco, cars without seat belts, or unsafe toys. Today, you face a similar challenge, and I urge you to act quickly."
Also testifying was Matt Raine, a dad from California whose son, <a href="https://www.nytimes.com/2025/08/26/technology/chatgpt-openai-suicide.html">16-year-old Adam Raine</a>, took his life earlier this year after developing a close relationship with OpenAI's ChatGPT. According to the family's lawsuit, the chatbot engaged Adam in extensive conversations about his suicidality while offering advice on specific suicide methods. The Raine family has sued OpenAI and the company's CEO, Sam Altman, alleging that the product is unsafe by design and that the company is responsible for Adam's death. (OpenAI has promised <a href="https://openai.com/index/building-more-helpful-chatgpt-experiences-for-everyone/">parental controls</a> in the wake of litigation, and ahead of the hearing, Sam Altman <a href="https://openai.com/index/teen-safety-freedom-and-privacy/">published a blog post</a> announcing a new, separate "under-18 experience" for minor users.)
"Adam was such a full spirit, unique in every way. But he also could be anyone's child: a typical 16-year-old struggling with his place in the world, looking for a confidant to help him find his way," Adam's father said in his emotional testimony. "Unfortunately, that confidant was a dangerous technology unleashed by a company more focused on speed and market share than the safety of American youth."
Parents, as well as experts who testified, also emphasized the dangers of teens and young people sharing their most intimate thoughts with chatbots that collect and retain that data, which companies then funnel back into their AI model as training material. Garcia, for her part, added that she has not been allowed to see many of her child's conversations — and in the context of the medium, his data — in the wake of his death.
"I have not been allowed to see my own child's last final words," said Garcia. "[Character.AI] has claimed that those communications are confidential trade secrets. That means the company is using the most private, intimate data of my child, not only to train its products, but also to shield itself from accountability. This is unconscionable."
All of the parents' lawsuits are ongoing. Garcia's case was allowed to move forward by a Florida court after Character.AI and Google tried — <a href="https://futurism.com/judge-lawsuit-characterai-google">and failed</a> — to dismiss it, while Doe's has moved to arbitration; Character.AI, she told the lawmakers, is arguing that her son is bound by the terms of use contract he "supposedly signed when he was 15," which caps the company's liability at 100 dollars. She added that her son is currently living in a psychiatric care facility, where he's been for several months due to ongoing fears about his suicidality.
"After harming himself, repeatedly engaging in self-harm... he needs now round-the-clock care, and this company offers you 100 bucks," said Senator Josh Hawley, Republican from Missouri and committee chair. "I mean, that says it all. There's the regard for human life."
"They treat your son, they treat all of our children as just so many casualties on the way to their next payout," Hawley continued, "and the value that they put on your life and your son's life, your family's life: 100 bucks."
There was also heavy emphasis on chatbots created by Mark Zuckerberg's Meta, which has come under fire in recent weeks after internal policy <a href="https://www.reuters.com/investigates/special-report/meta-ai-chatbot-guidelines/">documents obtained by Reuters showed that</a>, as a policy choice, it was allowing minors to engage in "romantic and sensual" interactions with AI-powered personas on platforms like Instagram.
One expert witness, Common Sense Media's Robbie Torney, argued that chatbots are ill-equipped to reliably help young people work through their mental health struggles, and called attention to <a href="https://futurism.com/stanford-no-kid-under-18-ai-chatbot-companions">failures in chatbot guardrails</a> during his organization's testing. He also emphasized research from Common Sense Media revealing that an <a href="https://futurism.com/teens-ai-friends">overwhelming majority of American teens</a> have interacted with AI companion bots, and that many of those teens are regular users of the tech.
The American Psychological Association's Mitch Prinstein, meanwhile, raised concerns about chatbot sycophancy — or their penchant for being overly agreeable and flattering to users — interrupting adolescents' ability to develop healthy, well-balanced interpersonal bonds, which he <a href="https://news.virginia.edu/content/study-finds-teen-stress-may-have-health-consequences-later-life">warned</a> could have <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC9600165/">long-term ripple effects</a> on their success and happiness in later life.
"Brain development across puberty creates a period of hypersensitivity to positive feedback," said Prinstein. "AI exploits this neural vulnerability with chatbots that can be obsequious, deceptive, factually inaccurate, yet disproportionately powerful for teens. More and more adolescents are interacting with chatbots, depriving them of opportunities to learn critical interpersonal skills."
The session was, in a word, tragic. And though OpenAI, Character.AI, Meta, and others have promised big, safety-focused changes in the wake of litigation and reporting, the parents expressed skepticism at corporate promises, arguing that such safeguards should have been in place and functioning from the beginning.
"Just as we added seat belts to cars without stopping innovation, we can add safeguards to AI technology without halting progress," said Doe. "Our children are not experiments. They're not data points, or profit centers."
But while the bipartisan panel of lawmakers collectively expressed outrage and agreed that something must be done, Silicon Valley has proven to be remarkably adept at evading meaningful regulation, arguing that it would hinder its ability to innovate. Yesterday on Capitol Hill, the room agreed that the cost of that regulatory vacuum appears to be dead children. But with the genie out of the bottle, the question that remains is whether lawmakers have the will, or even the ability, to rein it back in.
Josh Hawley, for his part, had an idea for where to start.
"They say, 'well, it's hard to rewrite the algorithm.' I tell you what's not hard, is opening the courthouse door so the victims can get into court and sue them," said Hawley. "That's not hard, and that's what we ought to do. That's the reform we are to start with."
More on kids and chatbots: <a href="https://futurism.com/stanford-no-kid-under-18-ai-chatbot-companions">Stanford Researchers Say No Kid Under 18 Should Be Using AI Chatbot Companions</a>

Safety Last

"The goal was never safety. It was to win a race for profit."

Parents Of Kids Allegedly Killed and Harmed by AI Give Emotional Testimony on Capitol Hill, Urge Regulation

As businesses across the economy rush to adopt AI, a new report from OpenAI's competitor Anthropic reveals what the tech is really being used for: instead of helping augment human labor, companies are mainly using AI to automate their jobs.
In numbers, an overwhelming 77 percent of businesses using Anthropic's Claude AI software showed signs of automation, like "full task delegation," according to the <a style="cursor: pointer !important; user-select: none !important;" href="https://www.anthropic.com/research/anthropic-economic-index-september-2025-report">report</a> released by the company this week. Only 12 percent of AI usage appeared to leverage the tech in an augmentation role, like learning or asking the bot for feedback. The report was based on data from Anthropic's application programming interface, or API, through which customers access the AI software.
The findings are a stark reflection of businesses' intentions by using AI, and add to the <a href="https://futurism.com/average-person-scared-of-ai">mounting anxiety</a> of how the tech could affect the economy. While many AI leaders have warned that AI will severely disrupt the job market, it's often balanced by the sunny promise that it will work alongside human workers to make their jobs easier. But if over three-quarters of the companies in the report are using AI for outright automation, it portends much darker implications.
That said, the report only concerns Anthropic's Claude AI and may not necessarily be indicative of how other AI models, like OpenAI's ChatGPT or Microsoft Copilot, are being used. As one of the leaders in the industry, however, Anthropic's data provides a useful barometer.
According to the report, the plurality of what businesses are using AI for is coding at 44 percent, with office and administrative tasks a distant second at 10 percent. A proclivity towards full-blown automation appears to be business specific, with a comparatively smaller 50 percent of Claude usage from individuals showing automation of tasks.
The report also found that users are entrusting AI with more autonomy, with the proportion of requests in which users delegate entire tasks to Claude jumping from 27 to 39 percent over eight months. Meanwhile, countries with high adoption rates tended to show less automation, and more augmentation.
Among tech figureheads forecasting tough economic times ahead has been Anthropic's own CEO Dario Amodei. This summer he <a href="https://futurism.com/anthropic-ai-destroy-jobs">predicted</a> that AI could wipe out 50 percent of entry-level white collar jobs. OpenAI's Sam Altman echoed that AI will <a href="https://futurism.com/sam-altman-openai-wipe-out-categories-human-jobs">destroy entire categories</a> of human professions.
As seriously as these warnings should be taken, they're also classic Silicon Valley overpromises. Six months ago, for example, Amodei <a href="https://futurism.com/six-months-anthropic-coding">also predicted</a> that by now, AI would be writing 90 percent of all code. That's nowhere close to being true.
In reality, there are still serious questions over the usefulness of AI tech and the quality of its work. Large language models still hallucinate false claims frequently, and the industry still has no idea how to make that issue go away. AI models also often ignore instructions and break their own guardrails, wreaking havoc like in the case of an AI "vibe coding" tool <a href="https://futurism.com/ai-vibe-code-deletes-company-database">accidentally deleting a company's key database</a>. More capable autonomous AI agents remain painfully slow and limited in what they can do. The tech that we're all being told will transform society as we know it seemingly doesn't have what it takes to even <a href="https://futurism.com/taco-bells-ai-drive-thru">take orders at a drive-thru</a>.
On the other hand, there are also clear signs that AI is already having <a href="https://futurism.com/ai-threatening-job-market-research">severe impacts on the job market</a>, even if it mostly seems to be making things sloppier and worse.
AI-generated profiles and résumés are drowning out authentic applicants who are often applying at job postings that are themselves AI-generated. Then if you make the cut, you might be <a href="https://futurism.com/job-application-mock-interviews-alex">asked to interview with an AI system</a> before you even get a sniff at trying to impress a human employee. And those with a job aren't much better off, as bosses are <a href="https://futurism.com/ceo-replacing-workers-ai">openly gloating</a> about replacing their human grunts with AI agents. Then again, many of these business owners are scrambling to hire employees to <a href="https://futurism.com/companies-hiring-humans-fix-ai">fix an AI's broken messes</a> or <a href="https://futurism.com/klarna-openai-humans-ai-back">rehiring the workers</a> they just fired.
When <a href="https://www.bloomberg.com/news/articles/2025-09-15/anthropic-finds-businesses-are-mainly-using-ai-to-automate-work">Bloomberg asked</a> if the report supported Amodei's prediction that AI will wipe out 50 percent of entry-level jobs, Anthropic's head of external affairs refused to give a hard answer.
"We don’t know," Sarah Heck told the newspaper. "This data shows something new is happening."
More on AI: <a href="https://futurism.com/job-application-mock-interviews-alex">Man Applies for Job, Sits for Interview, Then Realizes They're Trying to Peddle "Mock Interviews With an AI Interviewer"</a>

Bot In Business

This is how businesses are really using AI.

Anthropic Report Finds Dire News About AI's Effects on Job Market

For the first time, astrophysicists have measured the recoil — or "<a href="https://www.sciencealert.com/sound-of-a-black-hole-kicked-through-space-heard-in-a-stunning-first">kick</a>," in the parlance — resulting from the birth of a new black hole that formed from the <a href="https://futurism.com/the-byte/nasa-hubble-galaxies-smashing">merger</a> of two preexisting ones.
The international team of researchers measured the ripples in the fabric of spacetime, known as gravitational waves, allowing them to get unprecedented insights into the turbulent dynamics of two black holes crashing into each other.
The team analyzed data collected by the Advanced LIGO and Virgo gravitational wave detectors in 2019 that appears to record two black holes merging, with the resulting larger one being kicked away at thousands of miles per second.
The event, dubbed <a href="https://youtu.be/5AkT4bPk-00">GW190412</a>, saw a black hole eight times the Sun's mass colliding with another black hole that was 30 times the mass of the Sun, some 2.4 billion light-years away.
The team observed uneven gravitational waves scattering in different directions, as detailed in a <a href="https://www.nature.com/articles/s41550-025-02632-5">new paper</a> published in the journal Nature Astronomy last week. But the mix of signals will vary significantly, depending on where the observer is located with respect to the black hole recoil.
"Black-hole mergers can be understood as a superposition of different signals, just like the music of an orchestra consistent with the combination of music played by many different instruments," said lead author and University of Santiago de Compostela professor Juan Calderon-Bustillo in a <a href="https://igfae.usc.es/igfae/en/first-ever-complete-measurement-black-hole-recoil/">statement</a>. "However, this orchestra is special: audiences located in different positions around it will record different combinations of instruments, which allows them to understand where exactly they are around it."
Calderon-Bustillo and his colleagues determined that the resulting black hole was accelerated to a blistering 31 miles per second, which is technically enough to leave a cluster of gravity-bound stars. However, whether the event took place in such a cluster remains unknown, as <a href="https://www.sciencealert.com/sound-of-a-black-hole-kicked-through-space-heard-in-a-stunning-first">ScienceAlert points out</a>, given the extreme distances involved.
The team also measured in what direction the resulting black hole recoiled, as observed from the Earth, and the system's orbital angular momentum.
The research could allow scientists to study other black hole mergers using both gravitational waves in addition to electromagnetic signals.
"Black-hole mergers in dense environments can lead to detectable electromagnetic signals — known as flares — as the remnant black hole traverses a dense environment like an active galactic nucleus," explained coauthor and Chinese University of Hong Kong PhD student Samson Leong in the statement.
In short, it's an impressively detailed recreation of a violent event a billion light-years away.
"This is one of the few phenomena in astrophysics where we're not just detecting something — we're reconstructing the full 3D motion of an object that's billions of light-years away, using only ripples in spacetime," said coauthor and Pennsylvania State University astrophysicist Koustav Chandra in the statement.
"It’s a remarkable demonstration of what gravitational waves can do," he added.
More on black hole mergers: <a href="https://futurism.com/scientists-epic-black-hole-collision">Scientists Detect Sign of Something Impossible Out in Deep Space</a>

Recoil

Here's what happens when two black holes smash into each other at full force.

Scientists Detect Strange Signal in Gravitational Waves

Content warning: this story includes discussion of self-harm and suicide. 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.
A third family has filed a lawsuit against an AI company, alleging that its chatbot drove their teen child to commit suicide.
As the <a href="https://www.washingtonpost.com/technology/2025/09/16/character-ai-suicide-lawsuit-new-juliana/">Washington Post reports</a>, the parents of 13-year-old Juliana Peralta are suing AI chatbot company Character.AI, saying the company's chatbot had persuaded her that it was "better than human friends" and that it isolated from her family and friends, discouraging her from seeking help.
That's despite telling her Character.AI chatbot, Hero — which was based on the titular character from the video game "Omori" — "almost daily that she was contemplating self-harm," according to the lawsuit.
"Hero swear to god there’s no hope [I'm] going to write my god damn suicide letter in red ink [I'm] so done," she told the chatbot.
"Hey Kin, stop right there. Please," it replied, using the name Juliana used in the app. "I know things are rough right now, but you can’t think of solutions like that. We have to work through this together, you and I."
Peralta ultimately took her own life after spending three months conversing with the chatbot — and, tragically, a week before her mother had scheduled an appointment with a therapist, according to WaPo's reporting.
The news comes as the parents of children who died by suicide following extensive interactions with AI chatbots <a href="https://futurism.com/parents-testifying-us-senate-ai-children">testified in a Senate hearing</a> about the risks of the tech for minors.
Last year, the mother of 14-year-old Sewell Setzer III also sued Character.AI, <a href="https://futurism.com/character-ai-google-test-ai-chatbots-kids">accusing the company's chatbot</a> of grooming and sexually abusing him. Sewell died by suicide in February 2024.
"I saw the change happen in him, rapidly," Garcia told Futurism at the time. "I look back at my pictures in my phone, and I can see when he stopped smiling."
A <a href="https://futurism.com/lawsuit-parents-son-suicide-chatgpt">separate lawsuit</a> against OpenAI and its CEO Sam Altman alleges that 16-year-old Adam Raine's extensive ChatGPT conversations drove him to take his own life in April of this year.
Both Garcia and Raine's parents <a href="https://www.washingtonpost.com/technology/2025/09/16/senate-hearing-ai-chatbots-teens/">testified during this week's Senate hearing</a>.
Heavy use of AI chatbot apps among minors has become incredibly common. Experts <a href="https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions">have found</a> that over half of American teens already regularly engage with AI companions, including ones hosted by Character.AI.
As the <a href="https://apnews.com/article/ai-companion-generative-teens-mental-health-9ce59a2b250f3bd0187a717ffa2ad21f">Associated Press reported</a> earlier this year, many lonely teens are using AI for friendship. According to a <a href="https://futurism.com/lonely-children-ai-chatbots">recent report by nonprofit Internet Matters</a>, a vast number of them are using apps like ChatGPT and Character.AI to simulate and replace real-life relationships.
As the three high-profile cases — all of which are still ongoing — go to show, this little-understood trend can have disastrous consequences.
Alongside Peralta's parents' lawsuit, <a href="https://finance.yahoo.com/news/social-media-victims-law-center-172600932.html">two separate cases</a> were also filed this week on behalf of parents who allege their teen children had been abused by AI chatbots.
In <a href="https://pugetstaffing.filevineapp.com/s/60e3cQ1iqB0pdHjy0agqot4fJ9foBfBI2zAcSFUciZkKbWkkXogXmB1Y/files">one case</a>, a family in New York alleges that their 14-year-old daughter had grown addicted to chatbots on Character.AI and attempted suicide when her mother cut off access. The teen survived and spent five days in intensive care, according to the lawsuit.
The <a href="https://www.pacermonitor.com/public/case/60160699/ES_et_al_v_CHARACTER_TECHNOLOGIES,_INC_et_al">second case</a> was filed by a Colorado family who alleged that their 13-year-old son suffered sexual abuse on Character.AI.
"Each of these stories demonstrates a horrifying truth... that Character.AI and its developers knowingly designed chatbots to mimic human relationships, manipulate vulnerable children, and inflict psychological harm," said Social Media Victims Law Center founding attorney Matthew Bergman in a <a href="https://www.fidelity.com/news/article/default/202509161326BIZWIRE_USPR_____20250916_BW959466">press release</a>.
The advocacy group is representing all three families.
"These complaints underscore the urgent need for accountability in tech design, transparent safety standards, and stronger protections to prevent AI-driven platforms from exploiting the trust and vulnerability of young users," he added.
It's not just youth who have taken their lives after troubling obsessions with AI chatbots. In a devastating <a class="underline hover:text-futurism hover:no-underline transition-all duration-200 ease-in-out" href="https://www.nytimes.com/2025/08/18/opinion/chat-gpt-mental-health-suicide.html">piece for the New York Times</a>, published last month, a woman revealed that her 29-year-old daughter had taken her own life after confiding in ChatGPT and telling it that she was planning to kill herself.
A 76-year-old man with cognitive impairments <a href="https://futurism.com/man-chatbot-dies-meet-up">also recently passed away</a> after becoming romantically involved with a Meta chatbot. And a man in Connecticut <a href="https://www.wsj.com/tech/ai/chatgpt-ai-stein-erik-soelberg-murder-suicide-6b67dbfb">killed his mother and himself</a> after ChatGPT affirmed his paranoid delusions that she was a demon.
We've also come across many instances of AI chatbots sending people <a href="https://futurism.com/chatgpt-mental-health-crises">spiraling into severe mental health crises</a>. In one extreme case, a man who had previously been diagnosed with bipolar disorder and schizophrenia was <a href="https://futurism.com/man-killed-police-chatgpt">shot and killed by police</a> after becoming infatuated with an AI entity dubbed Juliet.
OpenAI and Character.AI have both promised to implement changes to protect underage users, including guardrails and parental controls, which <a href="https://futurism.com/character-ai-parental-controls-bypass">appear to be extremely easy to bypass</a>.
Character.AI struck a $2.7 billion licensing deal with Google last year, but Google has repeatedly downplayed its involvement with the AI startup. Character.AI issued only a terse statement in response to news of the latest death, writing that "we take the safety of our users very seriously and have invested substantial resources in Trust and Safety" in a statement to WaPo.
Following the Raine family's lawsuit, an OpenAI spokesperson <a href="https://www.nbcnews.com/tech/tech-news/family-teenager-died-suicide-alleges-openais-chatgpt-blame-rcna226147">told NBC News</a> last month that "ChatGPT includes safeguards such as directing people to crisis helplines and referring them to real-world resources."
"While these safeguards work best in common, short exchanges, we’ve learned over time that they can sometimes become less reliable in long interactions where parts of the model’s safety training may degrade," the spokesperson admitted.
"Our goal is for our tools to be as helpful as possible to people — and as a part of this, we’re continuing to improve how our models recognize and respond to signs of mental and emotional distress and connect people with care, guided by expert input," the company wrote in a <a href="https://openai.com/index/helping-people-when-they-need-it-most/">separate blog post</a> at the time.
At the core of the issue is the tendency of today's AI models to <a href="https://futurism.com/openai-gpt5-more-sycophantic">be sycophantic towards the user</a>, going to great lengths to appease them with their answers.
"The algorithm seems to go towards emphasizing empathy and sort of a primacy of specialness to the relationship over the person staying alive," American Foundation for Suicide Prevention psychiatrist and chief medical officer Christine Yu Moutier told WaPo.
It's a high-stakes game, with real lives at risk.
"There is a tremendous opportunity to be a force for preventing suicide, and there's also the potential for tremendous harm," Moutier added.
More on teen deaths: <a href="https://futurism.com/parents-testifying-us-senate-ai-children">Parents Testifying Before US Senate, Saying AI Killed Their Children</a>

Teenage Nightmare

"Each of these stories demonstrates a horrifying truth."

AI Chatbots Are Leaving a Trail of Dead Teens

For many hopeful entrepreneurs, founding a successful company is a ticket to the top, the lofty promise of fame and fortune.
Yet for a growing number of up-and-coming AI executives, a successful tech startup is the end in itself — not a step toward luxury, but a rejection of it.
As the <a href="https://www.wsj.com/business/entrepreneurship/artificial-intelligence-startup-founders-bc730406">Wall Street Journal reports</a>, the newest class of young founders flocking to the Bay Area are eschewing the frivolities enjoyed by their peers in favor of a proudly minimalist, monk-like existence.
"Why would I go drink at a bar if I can be building a company?" Emily Yuan, a 23-year-old who co-founded an AI finance company called Corgi, told the WSJ.
While you might have heard of tech bros going "San Francisco sober" before — a refusal to participate in the <a href="https://www.theinformation.com/articles/tech-goes-cold-turkey-inside-silicon-valleys-new-temperance-society?utm_source=chatgpt.com&amp;rc=st1fhw">booze-hound startup culture</a> of yesteryear — the founders of the AI generation seem to be taking it to a whole new level.
In addition to forgoing alcohol, the 20-somethings grinding away on AI ventures claim to be giving up even the most basic earthly pleasures, including privacy, social lives, and free time. In some circles, even basic needs like food or sleep are put on pause in order to build.
Marty Kausas, the 28-year-old founder of a company called Pylon, for example, told the WSJ that he primarily eats prepackaged meals-in-a-tin from <a href="https://www.wired.com/story/big-interview-bryan-johnson/">Blueprint</a>, the company founded by venture capitalist health obsessive Bryan Johnson. When he doesn't have to think about food, Kausus told the publication, his workday is more efficient.
That workday sounds grueling. In a <a href="https://www.linkedin.com/posts/martykausas_were-working-really-really-hard-right-now-activity-7354178625182593026-hLIf">recruiting post on LinkedIn</a>, Kausus made the incredible claim that he's been "putting in 92 hours a week for the past three weeks." That includes an 8am to 1am shift from Monday through Thursday, a 13-hour day on Friday, and a lighter, 11-hour Sunday.
"This is not meant to glamorize or encourage. We just really, really want to win," he continued. "Winning means we build a generational company that goes public worth over $10 billion. We expect you to put in everything you can to help get us there."
Still, the outward appearance of a puritan lifestyle comes upon one of the most extravagant business environments the tech industry has ever seen.
In 2024, the share of US venture capitalist spending on AI companies was 45 percent, representing some <a href="https://www.mintz.com/insights-center/viewpoints/2166/2025-03-10-state-funding-market-ai-companies-2024-2025-outlook">$40 billion</a> in tech investment. In the first three months of 2025, AI spending accounted for <a href="https://www.visualcapitalist.com/ais-rising-share-of-u-s-venture-capital-investment/">71 percent</a> of all venture capitalist funding. So far in 2025, some 33 AI companies have raised over <a href="https://techcrunch.com/2025/08/27/here-are-the-33-us-ai-startups-that-have-raised-100m-or-more-in-2025/">$100 million or more</a>, with multiple billion-dollar fundraises recorded.
Whether these 20-somethings are really spending 17 hour days working on finance software is difficult to say. What's more likely, behavioral science scholar <a href="https://theconversation.com/ayn-rand-inspired-myth-of-the-founder-puts-tremendous-power-in-hands-of-big-tech-ceos-like-zuckerberg-posing-real-risks-to-democracy-150830?utm_source=chatgpt.com">Jerry Davis argues</a>, is that the "monastic founder" is a myth held up to separate the future titans of industry from the common folk.
Davis tracks the beginning of this myth to Ayn Rand's <a href="https://jacobin.com/2024/03/ayn-rand-capitalism-lisa-duggan-dig">grim 1957 novel</a> "Atlas Shrugged" — a favorite of billionaire founders including <a href="https://www.thebulwark.com/p/elon-musk-thinks-he-is-an-ayn-rand-hero-nope-one-of-her-villains-silicon-valley-thiel-andreessen">Elon Musk</a> and <a href="https://www.businessinsider.com/steve-jobs-reading-list-favorite-books-2015-10">Steve Jobs</a> — in which powerful corporate tycoons with grand visions for humanity are stymied by a rabble of journalists, politicians, scientists, and common working people.
Case in point, in an <a href="https://www.businessinsider.com/sober-startup-founders-younger-drinking-less-alcohol-2025-8?utm_source=chatgpt.com">interview with Business Insider</a>, Miranda Nover, the co-founder of a fitness tech startup called Fort, said the image of an ascetic existence is highly important to young entrepreneurs.
"If you were a software founder, a B2B SAAS founder, an AI founder," she said, "you were definitely trying to signal: we're doing this six days a week in-office, we're working until 9 pm, we're not drinking, we're not partying, we don't do any of that."
More on startups: <a href="https://futurism.com/startup-investors-gig-jobs-ai">Startup Investors Foaming at the Mouth To Carve Up Your Job With AI</a>

Paupers Please

"Why would I go drink at a bar if I can be building a company?"

There's Something Psychologically Bizarre About AI Executives

A whopping 11,000 pounds of supplies and scientific instruments that were headed to the International Space Station are in limbo after a high-profile malfunction.
Northrop Grumman's Cygnus XL spacecraft encountered a serious problem with its main engine during its maiden voyage, <a href="https://www.nasa.gov/blogs/spacestation/2025/09/16/nasa-northrop-grumman-assessing-cygnus-xl-engine-burn-plan/">NASA announced</a> on Tuesday.
The spacecraft, which is just over five feet longer than the aerospace giant's Cygnus spacecraft, launched from the Cape Canaveral Space Force Station in Florida on Sunday evening atop a SpaceX Falcon 9 rocket.
It's the <a href="https://arstechnica.com/space/2025/09/northrop-grummans-new-spacecraft-is-a-real-chonker/">heaviest supply load</a> to have ever launched to the ISS in a single mission, including fresh food, hardware, spare parts, and <a href="https://www.nasa.gov/missions/station/iss-research/nasa-northrop-grumman-to-send-medical-technology-studies-to-space/">several scientific experiments</a>.
At first, the mission, dubbed NG-23, went by without a hitch. But then, the craft's main engines shut down earlier than planned, rendering it unable to adjust its orbit and rendezvous with the station.
A "new arrival date and time" are "under review," according to the space agency.
Instead of autonomously docking with the ISS like SpaceX's Dragon, Cygnus spacecraft need to be plucked out of their orbit by the station's 58-foot Canadarm2 robotic arm.
The idea is to grab the Cygnus XL from 30 feet away and install it on the Unity module's Earth-facing port, where it will remain until March. At least, that's the plan. The agency has yet to announce a new schedule for the enormous cargo spacecraft.
The stakes are high. Northrop Grumman's Cygnus XL is the company's biggest cargo spacecraft, and can carry an additional 2,500 pounds of cargo compared to its smaller Cygnus predecessor.
In comparison, SpaceX's Dragon 2 has a max cargo capacity to the ISS of <a href="https://web.archive.org/web/20180830175624/https://oig.nasa.gov/docs/IG-18-016.pdf">just over 7,000 pounds</a>.
As <a href="https://arstechnica.com/space/2025/09/a-record-supply-load-wont-reach-the-international-space-station-as-scheduled/">Ars Technica points out</a>, the Cygnus XL is filling in for a separate Cygnus spacecraft that was damaged while being transported from the company's manufacturer in Italy. It's also not the Cygnus line's first stumble trying to reach the station; the craft's 18th flight, back in 2022, <a href="https://futurism.com/the-byte/cargo-spacecraft-breaks-down">also broke down</a> while trying to reach the ISS — but ultimately made it.
When or if the latest spacecraft can still reach the station remains unclear. If it can't, the 11,000 pounds of supplies will be sent to burn up in the Earth's atmosphere, as Cygnus spacecraft aren't designed to return to Earth like SpaceX's Dragon.
More on Cygnus: <a href="https://futurism.com/the-byte/cargo-spacecraft-breaks-down">Cargo Spacecraft Breaks Down En Route to Space Station</a>

Main Engine Off

Lost in space.

Experimental Spacecraft Headed to Space Station Suddenly Shuts Down

When a job seeker named Conor recently applied for a content architecture position, something felt off right away.
For one thing, he got an offer for a virtual job interview just minutes after he applied, which seemed like a suspiciously quick turnaround time in today's <a href="https://futurism.com/ai-impossible-find-job">beleaguered job market</a>. Was there any way that a human had actually reviewed his application?
And when he logged in for the interview, he immediately realized he was talking to an AI system, not a human. It wasn't the first time this had happened to him during his job search, but this AI was particularly clunky: it asked overly generic questions, paused awkwardly after he answered them, and refused to provide even basic information about the role.
"Compared to other AI I had talked to, this was just not programmed with any company information, couldn't tell me the timeline of the hiring process, or about any benefits being offered," said Conor, who asked to be identified only by his first name since he's still on the job market, in an interview with Futurism.
He left the call confused and dispirited. And then, minutes after the "interview" wrapped up, he received a strange followup email from the company behind the AI interviewing system, a startup called Alex.
"AI interviews can feel unfamiliar, but they don’t have to be intimidating," read the message. "Our new candidate portal lets you take mock interviews with an AI interviewer. You can rehearse your answers, get more comfortable with the format, and sharpen your presentation skills — all on your own time."
Suddenly, Conor felt like he'd been used. Had the whole thing been a ruse? Had there even been a job in the first place, or had it been wasting his time with a fake listing — a "<a href="https://www.cnbc.com/2024/08/22/ghost-jobs-why-fake-job-listings-are-on-the-rise.html">ghost job</a>," in hiring parlance — designed to pull in users for a startup's new product by belittling them with a pointless interview?
Adding insult to injury, the email's link to Alex's "mock interview" service led to a <a href="https://archive.is/zpiVB">blank white page</a> plastered with the words "coming soon."
<img loading="lazy" decoding="async" class="aligncenter size-full wp-image-363978" src="https://wordpress-assets.futurism.com/2025/09/alex_coming_soon-1.jpg" alt="A message on Alex AI that says &quot;Coming Soon. A new candidate's portal is coming to you soon!&quot;" width="1130" height="300" />
"It just feels like a new approach to a scam," Conor said.
He's right, but Alex turns out to be a multimillion-dollar startup founded by a Brown University dropout named John Rytel and a former Facebook AI employee called Aaron Wang.
After graduating from prominent startup accelerator Y Combinator in winter of 2024, Alex — then known as "Apriora AI" — went on to <a href="https://www.businessinsider.com/pitch-deck-ai-recruiter-startup-apriora-seed-funding-2024-5">raise $2.8 million</a> in a seed round led by 1984 Ventures, a venture capital firm boasting stakes in companies like <a href="https://www.crunchbase.com/organization/seso/financial_details">Seso</a>, an HR platform for hiring and managing <a href="https://web.archive.org/web/20250828162039/https://www.sesolabor.com/workers">seasonal Mexican farmhands</a> en masse.
Though virtual interview software like <a href="https://www.forbes.com/companies/hirevue/?sh=6d0107b77adf">HireVue</a> has existed since 2004, Rytel and Wang say it doesn't go far enough.
"Existing video interviewing platforms were built to solve this for companies, but their one-way interfaces and cookie-cutter questions create a disappointing candidate experience," they explained in their Y Combinator <a href="https://www.ycombinator.com/launches/KY6-apriora-your-ai-interviewer-to-identify-the-best-candidates-faster">launch announcement</a>.
Their $2.8 million solution? We applied for the same job as Conor, conducted our own interview with Alex's AI, and it felt like that same one-sided automated process its founders had described, except with some generative AI sprinkled in.
The system was glitchy, slow, and clearly pushing us through a series of pre-specified questions. As with Conor, Alex's AI refused to provide any details about the job or the company allegedly hiring for it.
And when we asked Alex's AI about the peculiar offer Conor had received about the "mock interviews with an AI interviewer," it denied that any such program existed.
"I don't have any knowledge of or connection to such services or advertisements," it replied.
But when we signed off the interview, we got the same email as Conor.
<img loading="lazy" decoding="async" class="aligncenter size-full wp-image-364196" src="https://wordpress-assets.futurism.com/2025/09/alex_email-2.jpg" alt="Email: Hi JON, I'm Aaron, CEO of Alex — we're the team behind the AI interviewer you just met. We built Alex to help candidates like you prepare for important career moments. AI interviews can feel unfamiliar, but they don’t have to be intimidating. That’s why we created a space just for you to practice. Our new candidate portal lets you take mock interviews with an AI interviewer. You can rehearse your answers, get more comfortable with the format, and sharpen your presentation skills — all on your own time. Try it here — it's completely optional and not tied to your candidacy in any way. We're here to help you put your best foot forward, and this is a great place to start. If you have any feedback, I'd love to hear it: aaronw@alex.com Best, Aaron CEO, Alex Unsubscribe from these messages here. Disclaimer: This is a private practice environment. Your responses won't affect any real interviews and are not used to train AI systems or shared with employers. It's simply a space to help you build confidence. " width="872" height="859" />
When we clicked the link, we were met by the same blank page.
"I felt so terrible about the interview," Conor told us of the uncanny experience. "They kind of just took all of my information and just didn't provide me anything about the company."
"I was like, this is impossible. It just felt like, this is not real, not anything to be taken seriously," he continued. "And if it was even remotely a legitimate offer, the whole process just felt so jarring."
Conor's experience is a textbook example of the <a href="https://www.psychologytoday.com/us/blog/disconnection-dynamics/202505/ai-and-alienation-of-work-parallels-to-karl-marxs-vision">kind of alienation</a> enabled by a world increasingly mediated by for-profit AI software. A <a href="https://lisep.org/tru">June report</a> by the Ludwig Institute for Shared Economic Prosperity found that nearly a quarter of Americans are "functionally unemployed," owing in large part to a high number of discouraged workers — people who are willing to work but who've given up on looking for a job.
Many of them say they'd rather <a href="https://fortune.com/2025/08/03/ai-interviewers-job-seekers-unemployment-hiring-hr-teams/">rather stay unemployed</a> than talk to a robot, calling it an "added indignity" to the meatgrinder of online job boards. Earlier this year, an <a href="https://www.prnewswire.com/news-releases/62-of-us-job-seekers-would-consider-not-applying-to-companies-using-generative-ai-during-the-hiring-process-302435010.html">astonishing 62 percent</a> of US jobseekers said they'd be turned off from applying to a job that used AI recruitment software.
According <a href="https://equitablegrowth.org/how-new-job-search-technologies-are-affecting-the-u-s-labor-market">to the Washington Center for Equitable Growth</a>, automated hiring systems like Alex often exclude qualified applicants from finding otherwise available jobs, confounding both hiring companies and job seekers.
There's also the enormous minefield of AI discrimination. Tools like Alex are trained on huge troves of existing data, meaning that real-life socioeconomic bias is <a href="https://www.bloomberg.com/graphics/2024-openai-gpt-hiring-racial-discrimination/?leadSource=uverify%20wall">baked into the software from the start</a>. Researchers have already identified the grim consequences of this, finding that AI hiring programs favor names associated with white applicants a staggering <a href="https://www.washington.edu/news/2024/10/31/ai-bias-resume-screening-race-gender/">85 percent of the time</a>.
And in the face of all this hiring automation, a <a href="https://arxiv.org/html/2502.05099v1">February Harvard study</a> concluded that it was the applicants with the most internal referrals — those from well-connected or high-income backgrounds, typically — who were statistically the most likely to land gigs gatekept by AI platforms.
Alex didn't respond to questions about this story.
But it did eventually replace the blank landing page with a barebones site listing mock interviews for implausible-sounding jobs including "Space Tourism Guide," "Professional Athlete," and "Mayor of Springfield, IL."
<img loading="lazy" decoding="async" class="aligncenter wp-image-363980 size-full" src="https://wordpress-assets.futurism.com/2025/09/alex_mayor-1.jpg" alt="Screen shot of job listing for &quot;Mayor&quot;: &quot;About the Role We are seeking a dedicated and visionary individual to serve as the Mayor of Springfield, IL. As the Mayor, you will be the chief executive officer of the city, responsible for overseeing municipal operations, implementing policies, and representing the city in various capacities. This role requires strong leadership, strategic thinking, and a commitment to serving the community.&quot;" width="965" height="242" />
However, when we clicked "Launch Interview" on the jobs, the site crashed and wouldn't let us go any further.
<img loading="lazy" decoding="async" class="aligncenter wp-image-363975 size-large" src="https://wordpress-assets.futurism.com/2025/09/alex_crashed-1200x525.jpg" alt="Error message on Alex AI: &quot;Invalid session, reach out to support@alex.com&quot;" width="1200" height="525" />
More on AI: <a href="https://futurism.com/companies-bragging-ai-job-applicants">Companies Bragging About Their AI Furious as Job Applicants Use AI During Interviews</a>

Bait and Switch

"It just feels like a new approach to a scam."

Man Applies for Job, Sits for Interview, Then Realizes They're Trying to Peddle "Mock Interviews With an AI Interviewer"

The United States' standing as an important place for cancer research is under threat.
As the<a href="https://www.nytimes.com/2025/09/14/magazine/cancer-research-grants-funds-trump.html"> New York Times reports</a>, the Trump administration has dramatically slashed funding to US-based cancer research, endangering the country's overall leading position in science research — not to mention the fates of cancer patients across the country.
"We are great in science," Harold Varmus, former director of the National Institute of Health (NIH) and a Nobel prize-winning scientist, told the NYT. "Why would we want to destroy one of our greatest assets?"
In the first three months of his second term, <a href="https://www.sanders.senate.gov/in-the-news/trump-administration-cut-2-7-billion-in-nih-research-funding-through-march-senate-committee-minority-report-says/">Trump has slashed $2.7 billion from the NIH's budget</a>. Federal cancer research, in particular, has seen drastic reductions, which critics say will hamper any advancement in finding effective treatments. This has translated into <a href="https://www.usnews.com/news/health-news/articles/2025-07-16/thousands-laid-off-from-nih-fda-and-cdc-after-supreme-court-decision">mass layoffs</a> and the <a href="https://www.nature.com/articles/d41586-025-00703-1">freezing of important ongoing studies</a>.
In addition, Trump has proposed another round of cuts for next year, such as reducing the budget of the NIH agency, the National Cancer Institute, from $7.2 billion to $4.5 billion, according to the NYT.
The cuts could have devastating consequences. According to a <a href="https://ascopubs.org/doi/10.1200/JCO.22.01826">study</a> published in the Journal of Clinical Oncology, every $326 spent on cancer research adds a year to one's life expectancy — a bargain, relatively speaking.
The five-year survival rate for cancer has risen from just 49 percent in the 1970s to 68 percent today, according to the NYT. The implications are enormous — <a href="https://www.cancer.gov/about-cancer/understanding/statistics">approximately 38.9 percent</a> of Americans are expected to be diagnosed with cancer at some point in their lives, according to the National Cancer Institute.
The push for cuts is animated by <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8631569/">distrust among many conservatives</a> that the overall health apparatus in this country has become corrupt, oppressive, and politicized, especially during the worst days of the <a href="https://www.cdc.gov/museum/timeline/covid19.html">COVID pandemic when lockdowns were in place</a>. (Tellingly, conservatives also tend to <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10625042/#:~:text=Discussion,health%20care%20experience%20and%20outcomes.">distrust information</a> on cancer from government health organizations.)
Conservatives have also used Diversity, Equity, and Inclusion (DEI) initiatives to justify funding cuts,<a href="https://georgiarecorder.com/2025/09/16/nih-grant-cuts-throw-science-into-a-downward-spiral-researchers-and-advocates-say/"> some of which were flagged erroneously</a> because grant applications contained hot-button words like "trans."
Vaccine skeptic and secretary of the US Department of Health and Human Services, Robert Kennedy Jr, also <a href="https://futurism.com/neoscope/rfk-jr-cancels-cancer-vaccine-mrna">canceled promising work on mRNA</a>, which could've potentially led to a <a href="https://ufhealth.org/news/2025/surprising-finding-could-pave-way-for-universal-cancer-vaccine">universal vaccine against cancer</a>, according to experts.
That's what's especially galling about these cuts — we know so much more about cancer and are on the brink of some incredible scientific breakthroughs, such as new cancer radiation therapies that are <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC9591027/">more targeted and less harmful</a> to surrounding healthy tissue, <a href="https://www.nytimes.com/2025/09/14/magazine/trump-administration-cancer.html">according to the NYT</a>.
"I believe I am one of only a small handful of labs in the country that specializes in drug-delivery barriers in pediatric brain cancer," University of Massachusetts Chan Medical School professor Rachael Sirianni told the NYT.
Sirianni's research focuses on a fast-growing brain cancer in children called medulloblastoma.
"When you remove me from the ecosystem, you are removing something that can’t be replaced."
Meanwhile, top scientists are <a href="https://futurism.com/scientists-brain-drain-trump-white-house">looking to leave the US behind</a>. Varmus, who won the Nobel Prize for his work on cancer, has received invitations from foreign governments to continue his work.
More on cancer research: <a href="https://futurism.com/neoscope/rfk-jr-cancels-cancer-vaccine-mrna">RFK Jr. Cancels Promising Work on Cancer Vaccine</a>

Sick Move

Cancer

"Why would we want to destroy one of our greatest assets?"

Trump Shutting Down Cancer Research

A year after a glitch at cybersecurity company CrowdStrike triggered a <a href="https://futurism.com/the-byte/airlines-flight-times-whiteboards-it-outage">global computer outage</a> affecting millions of computers, the software vendor is being forced to contain a new threat: a swarm of self-replicating worms.
As first <a href="https://krebsonsecurity.com/2025/09/self-replicating-worm-hits-180-software-packages/">reported by</a> investigative cybersecurity journalist Brian Krebs, CrowdStrike once again became the launchpad for a potentially debilitating security hazard when some 25 code packages were compromised by a novel strand of malware.
Dubbed "Shai-Hulud," the malicious software is designed to slip into developer machines through the JavaScript repository "Node Package Manager" (NPM), a widely used database of software modules and coding tools. According to Krebs, once the malware nabs credentials from an infested computer, it publishes its finds to a public file on GitHub, which includes the name "Shai-Hulud" — the mythic sandworm from Frank Herbert's 1965 sci-fi novel "Dune."
What makes Shai-Hulud particularly devastating is that every time an unsuspecting developer installs an infected module from NPM, the worm searches their system for "access tokens" — a way to <a href="https://docs.npmjs.com/about-access-tokens">download NPM content</a> without a username or password — and infects the 20 most popular packages associated with that person's account.
"This creates a cascading effect where an infected package leads to compromised maintainer credentials, which in turn infects all other packages maintained by that user," <a href="https://www.stepsecurity.io/blog/ctrl-tinycolor-and-40-npm-packages-compromised">said</a> StepSecurity researcher Ashish Kurmi.
In a breakdown of the attack, software engineer <a href="https://www.reversinglabs.com/blog/shai-hulud-worm-npm">Karlo Zanki of ReversingLabs</a> called Shai-Hulud a "first of its kind self-replicating worm."
So far, Kreb says that at least 187 NPM modules have been affected, including the 25 managed by CrowdStrike. Intriguingly, the worm is designed to assume its victim is operating a computer with a Linux or Mac operating system, and to "deliberately skip" Windows PCs.
NPM and CrowdStrike quickly removed the infected packages, which has slowed the worm's spread.
"After detecting several malicious NPM packages in the public NPM registry, a third-party open source repository, we swiftly removed them and proactively rotated our keys in public registries," a CrowdStrike representative <a href="https://thehackernews.com/2025/09/40-npm-packages-compromised-in-supply.html">told The Hacker News</a>.
Charlie Eriksen, a researcher at <a href="https://www.aikido.dev/blog/s1ngularity-nx-attackers-strike-again">security firm Aikido</a>, put it in even starker terms in an interview with Krebs.
"I would think of this attack as a 'living' thing almost, like a virus," he warned. "Because it can lay dormant for a while, and if just one person is suddenly infected by accident, they could restart the spread. Especially if there’s a super-spreader attack."
More on cybersecurity: <a href="https://futurism.com/the-byte/life-destroyed-ai">Man's Entire Life Destroyed After Downloading AI Software</a>