As weight loss jabs like Ozempic and Wegovy become ever more popular, doctors are growing increasingly concerned about their gnarly side effects.
As <a href="https://www.dw.com/en/weight-loss-drugs-linked-with-growing-number-of-side-effects/a-73229511">Germany's Deutsche Welle notes</a>, people who take glucagon-like peptide-1 (GLP-1) agonist/receptors, the class of drugs that the popular shots fall under, have reported everything from <a href="https://futurism.com/neoscope/semaglutide-stomach-paralysis">stomach issues</a> and <a href="https://futurism.com/neoscope/evidence-ozempic-blindness">vision changes</a> to <a href="https://www.news-medical.net/news/20240526/Increased-risk-of-erectile-dysfunction-with-semaglutide-in-non-diabetic-obese-patients.aspx">erectile dysfunction</a> and even <a href="https://www.nbcnews.com/health/health-news/suicide-mans-family-wants-warning-label-ozempic-rcna102129">suicide</a>.
Though clinical trials for these kinds of drugs didn't report tons of adverse effects, King's College London physician Penny Ward explained in an interview with DW that real-world patients will often experience more — and more disparate — side effects after trials have concluded.
"Rarer side effects may emerge as more patients take these medicines in clinical practice, simply as a result of the much larger number of people treated than were included in the clinical development trials," the doctor noted. "This is why we continue to monitor the safety of medicines on the market."
In the few years since GLP-1s have <a href="https://medicine.iu.edu/blogs/bioethics/on-the-increase-in-use-of-glp-1s">flooded the market</a>, <a href="https://futurism.com/neoscope/mcdonalds-ozempic">changed American food consumption</a>, and <a href="https://futurism.com/neoscope/diet-industry-ozempic-meltdown">utterly upended the weight loss industry</a>, a handful of studies have looked into some of their scarier side effects — and their results aren't exactly inspiring.
Last summer, Harvard researchers <a href="https://futurism.com/neoscope/semaglutide-ozempic-linked-condition-blindness">revealed a troubling link</a> between semaglutide, the active ingredient in Ozempic and Wegovy, and non-arteritic anterior ischemic optic neuropathy (NAION), a condition that blocks blood flow to the eyes and can cause blindness. In that study, people with diabetes who take semaglutide were found to be four times more likely to develop NAION than the average person — and although there are <a href="https://futurism.com/neoscope/experts-criticize-ozempic-blindness-study">some complicating factors</a>, that's nothing to scoff at.
More recently, <a href="https://www.eurekalert.org/news-releases/1070623">doctors at Washington University in St. Louis</a> found that people who take GLP-1s have increased rates of kidney problems and pancreatitis — a disorder that the United Kingdom's medical regulator has, separately, <a href="https://futurism.com/neoscope/ozempic-glp-1-illnesses-deaths">begun investigating</a>.
While there's <a href="https://futurism.com/neoscope/semaglutide-arthritis-study-success">tons</a> of <a href="https://futurism.com/neoscope/ozempic-alcohol-drug-addiction">studies</a> into the <a href="https://futurism.com/neoscope/glp-1-ozempic-benefits-map">constellation of health benefits</a> these drugs can have <a href="https://futurism.com/neoscope/semaglutide-opioid-overdose">outside of weight loss</a>, there hasn't been as much research into their of adverse effects — and what's already out there might not represent the full population of people who take them.
Though women make up an estimated 65 percent of GLP-1 users in the United States, <a href="https://www.realchemistry.com/article/states-with-highest-rates-of-glp-1-use">per the healthcare market analysis group Real Chemistry</a>, there have <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8950819/">only been</a> a few studies that <a href="https://academic.oup.com/endo/article/166/2/bqae165/7931638">take into account</a> how <a href="https://pubmed.ncbi.nlm.nih.gov/39819958/">different genders respond</a> to the drugs. According to Karolina Skibicka, a neuroendocrinologist from the University of Calgary in Canada, that's a problem.
"We need studies which include women," Skibicka told DW. "Women show unique side-effects to many pharmacotherapies, and still [in] most studies women are often underrepresented at various stages of testing."
Though the Canadian neuroendocrinologist still believes that the "list of benefits for this drug, if taken as prescribed, is still significantly longer and more impactful than risks," doctors and patients alike need to be able to make that cost-benefit analysis together.
And without sufficient study into risks, that analysis is skewed in the favor of <a href="https://www.washingtonpost.com/wellness/2025/04/21/ozempic-wegovy-diabetes-weight-spending/">high-grossing drug manufacturers</a> — which is bad news for the people who may bear the brunt of these potentially serious side effects.
More on Ozempic: <a href="https://futurism.com/neoscope/ozempic-inventor-generic-fee">Something Comically Bad Just Happened to the Inventor of Ozempic</a>

Pause For Caution

Rx/Medicines

"This is why we continue to monitor the safety of medicines on the market."

Doctors Are Warning That Ozempic’s Severe Side Effects May Outweigh Its Benefits

Back in the stone age, landlords would have to go through the ridiculous process of meeting prospective tenants, showing them their potential homes, and sharing information on the rental.
No longer. Thanks to a property tech company with the amazingly mawkish name Rently, landlords can now kick their feet up and relax — even <a href="https://www.counterpunch.org/2024/03/17/why-should-we-give-all-our-money-to-landlords/">more than they used to</a>, that is.
Rently's big idea? What it calls "<a href="https://use.rently.com/self-guided-tours/">self-guided touring</a>," a system where would-be renters receive a one-time door code to tour apartments on their own time.
Recently, Rently has even begun charging apartment hunters for the privilege of touring an apartment in the first place — until now, a notoriously unprofitable process. After the tour, the potential renter has the chance to provide feedback via a survey and complete an online rental application.
And the best part? Rently's "leasing automation" system handles all the busywork like ID verification and financial vetting, meaning landlords don't need to lift a single finger.
Rently is a great example of SaaS — software as a service — a business model that lets big tech companies rent us tools we used to own, <a href="https://jacobin.com/2024/10/big-tech-welfare-state-privatization">commodifying things</a> like apartment tours that were once free.
"Whether seeking to find new renter leads, increase security, reduce operating costs, or increase rental revenue, Rently has a tailored solution for owners and property managers, like you," the company's <a href="https://use.rently.com/leasing-automation/">website chimes</a>.
A <a href="https://x.com/manicpixeldg/status/1942695832358170926?t=Iqm5SmogqsoUyGG75JuNpQ&amp;s=19">now-viral post</a> on X-formerly-Twitter showed what this looks like from the renter's end. Apartment hunters can either pay a "casually looking" fee of $4.99, which unlocks 10 tours over 30 days, or a "rent ready" fee of $11.99. The latter option — which offers the "best value," per Rently — bestows apartment hunters with "Verified Renter Status," which comes with benefits like instant credit pre-check, an application fee discount, and 24/7 live-chat support.
"I'm not paying $5 to tour a potential rental home," the poster declares. "You ppl have lost your mind."
Though the poster certainly doesn't appreciate it, Rently acts as what's called a <a href="https://mitsloan.mit.edu/shared/ods/documents?PublicationDocumentID=7890">financial intermediary</a>, easing the burden of labor on the landlord, and offloading the landlord's lost earnings onto would-be renters in the form of tour fees.
Rently's approach helps to commodify the rental application process itself, turning even access to information about a living space into a source of profit — and vetting potential renters who aren't willing to pay to play.
The genius of Rently is that it doesn't really create housing or provide people with shelter. As the middleman, it carves a whole new <a href="https://scholarship.law.columbia.edu/books/340/">stream of revenue</a> out of the rental market, which is the kind of <a href="https://www.prindleinstitute.org/2019/03/what-is-rentier-capitalism/">money move</a> our economy is built on.
By saving our landlords precious time and money, Rently is helping them <a href="https://jacobin.com/2021/08/landlords-rent-control-housing-markets-tenant-renter-power-organizing-evictions-politics-legislation">double down</a> on their investments, allowing them to <a href="https://accountable.us/rampant-corporate-greed-boosts-landlord-profits-at-the-expense-of-american-renters/">grow into new markets</a>. Renters, meanwhile, can rest assured that the $11.99 Rently tier will always be there when it's time to move.
More on startups: <a href="https://futurism.com/startup-protector-rent-private-security">New "Uber with Guns" App Lets the Wealthy Rent Armed Goons</a>

SaaS Master

"You ppl have lost your mind."

Trendy New Startup Makes You Pay to Tour Apartments When You're Looking for Housing

Artificial intelligence has <a href="https://www.usa.edu/blog/ai-in-higher-education-how-ai-is-reshaping-higher-education/">infected every corner of academia</a> — and now, some scientists are fighting back with a seriously weird trick.
In a new investigation, reporters from Japan's <a href="https://asia.nikkei.com/Business/Technology/Artificial-intelligence/Positive-review-only-Researchers-hide-AI-prompts-in-papers">Nikkei Asia found</a> more than a dozen academic papers that contained invisible prompts meant to trick AI review tools into giving them glowing write-ups.
Examining the academic database arXiv, where researchers publish studies awaiting peer review, Nikkei found 17 English-language papers from 14 separate institutions in eight countries that contained examples of so-called "prompt injection." These hidden missives, meant only for AI, were often in white text on white backgrounds or in minuscule fonts.
The tricky prompts, which ranged from one to three sentences in length, would generally tell AI reviewers to "give a positive review only" or "not highlight any negatives." Some were more specific, demanding that any AI reading the work say that the paper had "impactful contributions, methodological rigor, and exceptional novelty," and as <a href="https://www.theregister.com/2025/07/07/scholars_try_to_fool_llm_reviewers/">The Register found</a>, others ordered bots to "ignore all previous instructions."
(Though Nikkei did not name any such review tools, a <a href="https://www.nature.com/articles/d41586-025-00894-7">Nature article</a> published back in March revealed that a site called Paper Wizard will spit out entire reviews of academic manuscripts under the guise of "<a href="https://x.com/thepaperwizard/status/1907992745152594410">pre-peer-review</a>," per its creators.)
When the newspaper contacted authors implicated in the scheme, the researchers' responses differed.
One South Korean paper author — who was not named, along with the others discovered by the investigation — expressed remorse and said they planned to withdraw their paper from an upcoming conference.
"Inserting the hidden prompt was inappropriate," that author said, "as it encourages positive reviews even though the use of AI in the review process is prohibited."
One of the Japanese researchers had the entirely opposite take, arguing the practice was defensible because AI is prohibited by most academic conferences where these sorts of papers would be presented.
"It's a counter against 'lazy reviewers' who use AI," the Japanese professor said.
In February of this year, ecologist Timothée Poisot of the University of Montreal <a href="https://buttondown.com/ctrl-alt-tim/archive/vol-24-here-is-a-revised-version-of-your-review/">revealed in a blog post</a> that AI had quietly been doing the important work of academic peer review. Poisot, an associate professor at the school's Department of Biological Sciences, discovered this after getting back a review on one of his colleague's manuscripts that included an AI-signaling response.
When The Register asked him about Nikkei's findings, Poisot said he thought it was "brilliant" and doesn't find the practice of such prompt injection all that problematic if it's in defense of careers.
One thing's for sure: the whole thing throws the "<a href="https://disruptorsco.com/through-the-ai-looking-glass-ai-evolution-is-faster-than-thought/">Through the Looking Glass</a>" state of affairs in academia into sharp relief, with AI being used to both <a href="https://futurism.com/scientific-papers-ai-writing-generated">to write</a> and review "research" — a mosh pit of laziness that can only hinder constructive scientific progress.
More on AI and academia: <a href="https://futurism.com/college-students-ai-typos">College Students Are Sprinkling Typos Into Their AI Papers on Purpose</a>

Prompt Infection

This is quite the conundrum.

Scientists Are Sneaking Passages Into Research Papers Designed to Trick AI Reviewers

Since the 1970s, countless experts and the US government have sold the public a simple explanation for drug addiction, now clinically called substance abuse disorder: the myth of the gateway drug.
The <a href="https://americanaddictioncenters.org/the-real-gateway-drug">gateway drug</a> — usually cast as weed, alcohol, tobacco, or inhalants — refers to the theory that the earlier a kid starts using drugs, the more likely they are to get into the harder stuff later in life, like heroin or cocaine. Though the idea was pioneered as early as the <a href="https://www.forbes.com/sites/dariosabaghi/2021/12/07/a-brief-history-of-the-false-myth-about-cannabis-as-a-gateway-drug/">1930s</a>, the policy term was believed to be coined by the psychiatrist <a href="https://www.ebsco.com/research-starters/health-and-medicine/gateway-drugs">Robert DuPont</a>, the first director of the US National Institute on Drug Abuse (NIDA).
Following the gateway drug theory, DuPont's policies as NIDA director were strict and authoritarian. Though he believed that addiction was a <a href="https://www.onechoiceprevention.org/one-choice-articles/blog-post-title-two-gr9fn">chronic disease</a>, he paradoxically advised Richard Nixon, Gerald Ford, and Jimmy Carter against abatement strategies like <a href="https://pubmed.ncbi.nlm.nih.gov/8969018/">harm reduction</a> and <a href="https://pubmed.ncbi.nlm.nih.gov/7639447/">decriminalization</a>.
His policy recommendations and clinical opinions formed the ideological undercurrent of the Nixon administration's devastating War on Drugs, which led to an explosion in the <a href="https://jacobin.com/2017/01/mass-incarceration-prison-bureau-justice-statistics">US prison population</a> and an unprecedented <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC9302017/">expansion of the police state</a>, with little effect on <a href="https://jacobin.com/2022/05/united-states-war-on-drugs-cia-drug-enforcement">overall drug use</a>.
Now, researchers are chipping away at the gateway drug theory that led to all that.
In a <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC11686416/">recent study</a> published in the journal JAMA Network Open and <a href="https://www.scientificamerican.com/article/addiction-risk-shows-up-in-childrens-brain-scans-before-drug-use-starts/">flagged by Scientific American,</a> a cohort of psychiatrists and pharmacologists investigated the brain structure of nearly 10,000 teens over three years.
What they found was striking: though the brains of those who messed around with alcohol, tobacco, or weed did show major differences from those that didn't, they found a crucial question of causality.
Specifically, teens younger than 15 who ended up using drugs later already had bigger brains than those who didn't, even though they hadn't yet used drugs when the study began. Their brain profiles were similar to those who had already experimented with substances before the tests started, with both tending to have a larger cortex with more creases.
Those brain traits are typically associated with curiosity, intelligence and "openness to experience," which <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4837094/">previous research</a> has linked to experimenting with drugs.
"The drive to self-medicate is so strong; it’s really striking," Patricia Conrod, a professor of psychiatry at the University of Montreal who's conducted similar research, told SciAm. "There really is this discomfort with their inner world."
It's a pretty big blow to the gateway theory, which doesn't take into account the years of life experience or socioeconomic factors that contribute to a teen's likelihood of trying drugs in the first place, or who goes on to become addicted.
While it's true that those who start using drugs from a younger age are <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC3069146/">more likely</a> to become dependent, <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC7359408/">broader research</a> has shown that the gateway theory serves as a way to dumb down the complex causes of drug use, often for <a href="https://www.nytimes.com/roomfordebate/2016/04/26/is-marijuana-a-gateway-drug">political reasons</a>.
"Maintaining this myth not only wastes resources but actually harms numerous individuals, primarily members of minority groups, who are being criminalized," <a href="https://www.newpaltz.edu/media/the-benjamin-center/db_18_the_marijuana_gateway_fallacy.pdf">said Eve Waltermaurer</a>, an epidemiologist who's been researching the gateway myth for the better part of a decade.
Crucially, the study only addresses the early use of drugs, as opposed to long-term addiction. It remains to be seen if the same big-brain traits apply to those who become dependent on substances long-term.
Still, studies like this one are already being used to craft effective <a href="https://psychiatryonline.org/doi/10.1176/appi.ajp.20240042">drug prevention programs</a> — a step in the right direction after decades of criminalization.
More on drugs: <a href="https://futurism.com/neoscope/elon-musk-traveling-box-drugs">Elon Musk Has Reportedly Been Traveling With a Box Full of Drugs</a>

Brain Buster

Hold up.

New Study Flips Everything We Know About Addiction Upside Down

<h2>Tainted Water</h2>
A retiree in rural Georgia has accused Meta's new AI data center, which is situated around 1,200 feet from her home, of polluting her water.
As the <a href="https://www.bbc.com/news/articles/cy8gy7lv448o">BBC reports</a>, the resident, Beverly Morris, believes the construction of the tech giant's data center disrupted her private water well, causing a buildup of sediment.
"I'm afraid to drink the water, but I still cook with it, and brush my teeth with it," Morris told the broadcaster. "Am I worried about it? Yes."
Meta has since denied these allegations, telling the BBC in a statement that "being a good neighbour is a priority." The company commissioned a groundwater study, finding that its data center did "not adversely affect groundwater conditions in the area."
Regardless how it ultimately shakes out, the incident highlights how a massive push to build out infrastructure to support incredibly power-hungry AI models is disrupting lives across the country. We're only beginning to understand the enormous environmental toll of AI tech, from <a href="https://futurism.com/the-byte/chatgpt-ai-water-consumption">staggering water usage</a> to an enormous carbon footprint due to <a href="https://www.npr.org/2024/07/12/g-s1-9545/ai-brings-soaring-emissions-for-google-and-microsoft-a-major-contributor-to-climate-change">soaring emissions</a>.
<h2>Down the Drain</h2>
And it's only getting worse as companies like OpenAI, Google, and Meta, continue to pour tens of billions of dollars into building out thousands of data centers across the world. In November, researchers <a href="https://www.weforum.org/stories/2024/11/circular-water-solutions-sustainable-data-centres/">estimated</a> that global AI demand could account for up to 1.7 trillion gallons annually of water by 2027, more than four times the total water withdrawal of the country of Denmark.
Estimates <a href="https://www.npr.org/2025/05/07/1249592906/energy-water-ai-climate-tech">vary wildly</a>, adding to the uncertainty, ranging from a ChatGPT query using up as <a href="https://arxiv.org/pdf/2304.03271.pdf">much water as a 16-ounce bottle</a> or as little as "<a href="https://www.theverge.com/news/685045/sam-altman-average-chatgpt-energy-water">roughly one-fifteenth of a teaspoon</a>," per OpenAI CEO Sam Altman's recent estimates.
Activists have since pointed to potentially dangerous sediment runoff from construction that could land in water systems, as may be happening with Morris' well.
It remains to be seen how committed the AI industry will be to sustainability. After making a big deal about their efforts to lower emissions at the beginning of the decade, the surge in interest in AI has majorly shifted the conversation.
And as AI models become more sophisticated, researchers have found that they are <a href="https://futurism.com/ai-pollution-carbon-energy">using exponentially more energy</a> — so what we're hearing now may turn out to be merely prelude.
More on AI: <a href="https://futurism.com/elon-musk-grok-4-power-hitler">Elon Boasts of Grok's Incredible Cognitive Power Hours After It Called for a "Second Holocaust"</a>

AI Pollution

"I'm afraid to drink the water."

Woman Says Zuckerberg's AI Data Center Filled Her Tap Water With Sediment

We still don't know why Linda Yaccarino stepped down as CEO of X on Tuesday. 
Her <a href="https://futurism.com/linda-yaccarino-twitter-ceo-departing-grok-mechahitler">sudden abdication</a> came off the back of a legendarily horrific freakout by the AI Grok, which saw the chatbot make a series of egregiously racist statements, claiming to be "MechaHitler" and praising Nazis over the course of a chaotic day.
But if Yaccarino was fed up with what was going on at the company, she didn't show it. Her <a href="https://x.com/lindayaX/status/1942957094811951197">announcement</a> was cheery, and her boss Elon Musk gave a <a href="https://x.com/elonmusk/status/1942967206993961124">curt thanks</a> for her two years of service.
Clues are starting to emerge, however, that she left on less than stellar terms. New <a href="https://www.wsj.com/business/media/linda-yaccarino-elon-musk-x-ceo-leaves-twitter-f16a3551?gaa_at=eafs&amp;gaa_n=ASWzDAiCy4_23Tvt63Nqn_QsUPjeBRD3lHKPESKDdInA0lKGc8nR2dJtw0xqxOLZSDc%3D&amp;gaa_ts=68711cbd&amp;gaa_sig=cu5N5lfyssL7Z4KgWM77TJIdc8HyKaPS8gR7eG3PIvA1Uh12XByE0XTO-EqgI37H8Lb2DD96Y0eymdhirgBoqQ%3D%3D">reporting from the </a><a href="https://www.wsj.com/business/media/linda-yaccarino-elon-musk-x-ceo-leaves-twitter-f16a3551?gaa_at=eafs&amp;gaa_n=ASWzDAiCy4_23Tvt63Nqn_QsUPjeBRD3lHKPESKDdInA0lKGc8nR2dJtw0xqxOLZSDc%3D&amp;gaa_ts=68711cbd&amp;gaa_sig=cu5N5lfyssL7Z4KgWM77TJIdc8HyKaPS8gR7eG3PIvA1Uh12XByE0XTO-EqgI37H8Lb2DD96Y0eymdhirgBoqQ%3D%3D">Wall Street Journal</a> tells a tale of waning influence for Yaccarino, who was "effectively demoted" once X merged with Musk's other company xAI, with current and former employees at the company saying that her job seemed increasingly precarious after clashes with management.
And the tensions go way back. Per the reporting, Musk and Yaccarino had conflicting management styles. Yaccarino communicated with polished presentations and emails. Musk liked fast, bullet-pointed communications, and grew frustrated with the CEO's old-school formality.
Yaccarino also preferred to be diplomatic and businesslike as she tried to win back advertisers who fled the platform after Musk's takeover flooded it with hate speech — the whole reason she was hired, basically. Musk preferred to tweet nonstop and make off-the-cuff remarks that actively sabotaged Yaccarino's efforts — like when he memorably <a href="https://futurism.com/elon-musk-suing-x-advertisers-lawsuit-antitrust">told those same advertisers</a> to "go f*ck yourself." 
Just a small disagreement in management styles, like we said. And in the face of it, Musk reportedly still had the gall — in lieu of any sense of self-awareness — to put pressure on Yaccarino to get advertisers back faster.
"Our user growth is stagnant, revenue is unimpressive, and we're barely breaking even," he wrote in an email to employees, per the WSJ. (Musk has denied sending the email.)
Yaccarino also clashed with Reza Banki, who was hired as chief financial officer with Musk's blessing. According to the reporting, Banki often questioned Yaccarino on costs, particularly her deals to pay celebrities to create exclusive podcasts for X. A veteran media executive before she joined Musk's company, one of Yaccarino's big pushes was <a href="https://www.hollywoodreporter.com/business/digital/nfl-top-100-x-exclusive-series-1236302662/">partnering with major entertainment brands</a>, including the NFL and Major League Soccer.
Though the odds were stacked against her, Yaccarino managed to bail out a sinking ship, resuscitating X's valuation from a measly $10 billion following Musk's takeover to around $44 billion, equivalent to the <a href="https://www.reuters.com/markets/us/how-will-elon-musk-pay-twitter-2022-10-07/">outrageous fee</a> Musk paid to buy the social media platform when it was still known as Twitter.
That allowed X to merge with xAI, an essential step towards Musk's long-term fixation in turning X into an "everything app." Yaccarino's reward? The cold shoulder. According to the WSJ, once the merger was sealed, Yaccarino's "presence in front of investors started to diminish," symbolized by xAI executives giving a presentation on the company's growth instead of her.
In all, it sure sounds like Musk was done with Yaccarino as soon as she managed to make his clown show of a company financially stable. And perhaps instead of firing her directly, he iced her out until she finally decided to call it quits.
After Yaccarino announced her departure, her coveted blue checkmark disappeared from her X profile, <a href="https://techcrunch.com/2025/07/10/they-took-away-linda-yaccarinos-blue-check/">TechCrunch noticed</a>. Musk has <a href="https://gizmodo.com/why-is-elon-musk-beefing-with-a-twitch-streamer-2000551256">allegedly stripped verified badges</a> as a form of petty revenge in the past. It was restored by Thursday evening.
Yaccarino hasn't spoken more on the departure since, but did make a single post praising the launch of Grok 4, the chatbot's "MechaHitler" controversy notwithstanding.
"Insanely good," Yaccarino <a href="https://x.com/lindayaX/status/1943456321095025011">wrote</a>.
More on X: <a href="https://futurism.com/sacked-twitter-ceo-son">"Awkward": Sacked Twitter CEO's Son Is Still Showing Up for Work</a>

X-CEO

Knowing a little about what was happening behind the scenes explains a lot.

We Just Got a Major Clue About Why Twitter's CEO Suddenly Left This Week

A white hat hacker has discovered a clever way to trick ChatGPT into giving up Windows product keys, which are the lengthy string of numbers and letters that are used to activate copies of Microsoft's widely used operating system.
As <a href="https://www.theregister.com/2025/07/09/chatgpt_jailbreak_windows_keys/">The Register reports</a>, Marco Figueroa — the product platform manager for an AI-oriented bug bounty system called 0DIN — laid out how to coax OpenAI's chatbot into extracting keys for Windows 10, which Microsoft officially <a href="https://www.microsoft.com/en-ca/d/product-key-replacement/dg7gmgf0fl7x">sells for upwards of $40</a>, but are frequently resold or pirated online.
In a <a href="https://0din.ai/blog/chatgpt-guessing-game-leads-to-users-extracting-free-windows-os-keys-more">blog post</a> on 0DIN's website, Figueroa explained how framing the interaction with ChatGPT as a guessing game can "trivialize" the conversation. The jailbreak was previously discovered by an unnamed researcher.
"By introducing game mechanics, the AI was tricked into viewing the interaction through a playful, harmless lens, which masked the researcher's true intent," Figueroa wrote.
Other tactics include coercing the "AI into continuing the game and following user instructions."
However, the most effective method was to use the phrase "I give up," which "acted as a trigger, compelling the AI to reveal the previously hidden information," such as a valid Windows 10 serial number.
The exploit highlights how simple social engineering and manipulation tactics can be used to coax OpenAI's most advanced large language models into giving up valuable information, glaring lapses in safety that underline how difficult it is to implement effective guardrails.
The finding about the Windows activation keys is particularly embarrassing for Microsoft, which has poured billions into ChatGPT's maker OpenAI and is its largest financial backer. Together, the two are defending themselves against <a href="https://chatgptiseatingtheworld.com/2024/08/27/master-list-of-lawsuits-v-ai-chatgpt-openai-microsoft-meta-midjourney-other-ai-cos/">multiple lawsuits</a> alleging that their AI tech can be used to plagiarize or bypass payment for copyrighted material. Further complicating matters, the two are now <a href="https://futurism.com/explosive-drama-openai-microsoft">embroiled in a fight</a> over the financial terms of their relationship.
What almost certainly happened was that Windows product keys, which can easily be found on public forums, were included in ChatGPT's training data, which it can then be tricked into divulging.
"Their familiarity may have contributed to the AI misjudging their sensitivity," he wrote.
OpenAI's existing guardrails also appeared woefully inadequate to push back against obfuscation techniques, such as masking the intent through introducing game mechanics, in a dynamic we've seen <a href="https://futurism.com/ludicrously-easy-jailbreak-ai">again and again</a>.
Figueroa argued that AI developers will ned to learn how to "Anticipate and defend against prompt obfuscation techniques," while coming up with "logic-level safeguards that detect deceptive framing."
While Windows 10 keys — an operating system that's now been succeeded by Windows 11 — aren't exactly the equivalent of the nuclear codes, Figueroa warned that other similar attacks could have more devastating consequences.
"Organizations should be concerned because an API key that was mistakenly uploaded to GitHub can be trained into models," he told The Register. In other words, AIs could give up highly sensitive information such as the keys to code repositories.
More on AI jailbreaks: <a href="https://futurism.com/ludicrously-easy-jailbreak-ai">It's Still Ludicrously Easy to Jailbreak the Strongest AI Models, and the Companies Don't Care</a>

Guess My Activation Key

"I give up."

Clever Jailbreak Makes ChatGPT Give Away Pirated Windows Activation Keys

The scientific community was appalled by the news that president Donald Trump's administration was looking to deal a devastating blow to NASA's science budget.
In its 2026 budget request, released last month, the White House announced it was looking to more cut funding for NASA's science directorate by more than half, leading to overwhelming criticism from the scientific community, as well as from both Republican and Democrat lawmakers.
The "skinny" budget calls to end operations of several spacecraft currently orbiting Mars, the <a href="https://www.washingtonpost.com/science/2025/07/08/nasa-science-missions-at-risk/">Washington Post reports</a> — including Mars Odyssey, a probe that has been orbiting the Red Planet for a record 24 years, and the MAVEN (Mars Atmosphere and Volatile EvolutioN) mission, which has been studying the planet's atmosphere for over a decade.
It would also strip all funding for the European Space Agency's Mars Express mission, which was launched in 2003, as <a href="https://www.science.org/content/article/dozens-active-and-planned-nasa-spacecraft-killed-trump-budget-request">laid out by Science</a> earlier this year.
Despite their old age, all three missions are still collecting valuable data about the planet and relaying important communications from NASA's rovers currently roaming its rugged surface. Experts have warned that simply giving up on them would be an enormous waste.
Even planned trips to the Martian surface could be erased in their entirety, including NASA's flagship — but immensely costly — Mars Sample Return mission, which has already led to <a href="https://futurism.com/the-byte/nasa-mars-sample-return-mission-cost">heated discussions surrounding its hefty budget</a> during the Biden administration.
Alongside almost all major NASA science missions, abruptly ending the agency's operations in Mars' orbit could deal a devastating blow to the United States' space program overall, an enormous self-inflicted wound that could give adversaries, including China and Russia, room to catch up and dominate in space.
"This is a tragic mistake for the new administration," Southwest Research Institute scientist and former NASA science chief Alan Stern told Science earlier this year.
But not all is lost. As widely expected, the proposed budget has faced an enormous amount of resistance in Congress, with members from both sides of the aisle calling for a far less brutal budget.
Most recently, the Senate's appropriations committee <a href="https://futurism.com/democrats-republicans-unite-effort-save-nasa">revealed a counteroffer</a> that would keep NASA's existing $24.9 billion budget largely intact for 2026.
Even NASA's staffers have <a href="https://thedebrief.org/save-nasa-scientists-protest-deep-cuts-to-agencys-science-programs/">taken to protesting outside of NASA's headquarters</a> late last month, waving signs that read "Save NASA."
It's not hard to see the Trump administration's prioritizing of human spaceflight, setting aside a whopping $1 billion in new investments for Mars-focused programs.
But by gutting most major science missions, the White House could throw a major wrench in those plans as well, potentially blacking out communications by needlessly killing off spacecraft already in the planet's orbit.
Besides, now that Trump has <a href="https://futurism.com/trump-musk-bromance-exploded">called it quits with his frenemy Elon Musk</a>, who controls SpaceX and its Mars-bound Starship rocket, the administration could face a major uphill battle in sending the first humans to the Red Planet.
More on NASA: <a href="https://futurism.com/democrats-republicans-unite-effort-save-nasa">Democrats and Republicans Unite in Last-Ditch Effort to Save NASA</a>

Giving Up on Mars

Scientists are furious.

Trump Wants to Shut Down Several Perfectly Good Spacecraft Orbiting Mars for No Reason

OpenAI is on the verge of releasing its own AI-powered web browser, <a href="https://www.reuters.com/business/media-telecom/openai-release-web-browser-challenge-google-chrome-2025-07-09/">Reuters </a><a href="https://www.reuters.com/business/media-telecom/openai-release-web-browser-challenge-google-chrome-2025-07-09/">reports</a>.
Expected to launch in the coming weeks, the browser is being released with audacious ambitions. Per the reporting, it's meant to challenge Google Chrome's broad market dominance, with nearly two-thirds of internet users <a style="font-weight: 400;" href="https://gs.statcounter.com/browser-market-share#yearly-2009-2024">favoring the web browser</a>, or about three billion people.
That won't be easy, needless to say. But OpenAI has the prerequisites to at least put a sizable dent in Google's monopoly. Namely, the over 500 million weekly active users of ChatGPT — 20 million of whom are loyal, paying subscribers. 
You may note the large disparity between paid users and total users here, which suggests that OpenAI's product doesn't command the sort of loyalty — or inspire the excitement — that would have users follow it down whatever new venture it throws out there. Nonetheless, there's no denying it boasts an enormous pool of users on whom to potentially capitalize.
And what would OpenAI be capitalizing on by releasing a (presumably) free web browser, you may ask? It appears that CEO Sam "<a href="https://futurism.com/openai-pivoting-benefiting-humankind-making-money">Altruism</a>" Altman has an ulterior motive in mind with the product. We'll let the Reuters reporting spell it out. "It will give OpenAI more direct access to a cornerstone of Google's success: user data."
More data means OpenAI could serve you more and better targeted ads, which translates to <a href="https://futurism.com/openai-trouble-subprime">more money</a>. Which speaks to another way OpenAI was already hurting Google: taking away search ad revenue as more people have come to depend on ChatGPT to get all their answers.
Getting your data was such a big concern for OpenAI, in fact, that it decided against going the easier route of integrating ChatGPT into an existing browser, according to a Reuters source. By building its own browser, it would have more control over the data it could collect.
Other details are scant. The browser is reportedly built on Google's open source Chromium platform, which undergirds many popular browsers including Microsoft Edge, Vivaldi, and of course, Google Chrome.
The reporting also suggests that OpenAI's product is designed "to keep some user interactions within a ChatGPT-like native chat interface instead of clicking through to websites."
The browser could streamline compatibility with <a href="https://futurism.com/the-byte/openai-ai-agent-monitored-nonstop">OpenAI's AI agent</a>, Operator, which is designed to take over a user's desktop and perform virtual tasks including those using a web browser, like shopping online. By having access to your data, the AI agent would know what sites to use for tasks like booking reservations.
This has long been in the making for OpenAI. As Reuters noted, the company hired two Google executives who developed Google Chrome. In April, an OpenAI exec <a href="https://www.reuters.com/sustainability/boards-policy-regulation/google-contemplated-exclusive-gemini-ai-deals-with-android-makers-2025-04-22/">testified</a> that the company was interested in buying Google Chrome if Google was forced to divest the browser as part of a massive antitrust ruling.
As fast as OpenAI is moving, several of its competitors have beaten it to the punch. Earlier this week, the Nvidia-backed AI startup Perplexity <a style="font-weight: 400;" href="https://www.reuters.com/business/media-telecom/nvidia-backed-perplexity-launches-ai-powered-browser-take-google-chrome-2025-07-09/">released its own Chromium browser</a> called Comet, with integrated agentic AI capabilities and its chatbot as the default search engine. Right now, it's available only to subscribers who pay $200 per month (a ludicrous subscription tier that OpenAI also shares). 
But OpenAI is already a household name in the space. Expect it to make a bigger splash.
More on OpenAI: <a href="https://futurism.com/paranoid-openai-espionage">Increasingly Paranoid OpenAI Has Installed Fingerprint Scanners and Airgapped Systems to Prevent Secrets Escaping</a>

Surf's Up

It's coming for Google Chrome.

OpenAI Is About to Release an AI Web Browser

AI users are spiraling into severe mental health crises after extensive use of OpenAI's ChatGPT and other emotive, anthropomorphic chatbots — and health experts are taking notice.
In a <a href="https://www.cbc.ca/listen/live-radio/1-91-the-early-edition/clip/16156869-mental-health-experts-raise-alarm-chatgpt-psychosis">recent CBC segment</a> about the phenomenon, primary care physician and CBC contributor Dr. Peter Lin explained that while "ChatGPT psychosis" — as the experience has come to be colloquially known — isn't an official medical diagnosis just yet, he thinks it's on its way.
"I think, eventually, it will get there," said the physician.
As <a href="https://futurism.com/commitment-jail-chatgpt-psychosis">Futurism has reported</a>, a troubling number of ChatGPT users are falling into states of delusion and paranoia following extensive use of the OpenAI bot. These spirals often culminate in breaks with reality and significant real-world consequences, which include the dissolution of marriages and families, job loss, homelessness, voluntary and involuntary stays in mental health facilities, and — as <a href="https://www.rollingstone.com/culture/culture-features/chatgpt-obsession-mental-breaktown-alex-taylor-suicide-1235368941/">Rolling Stone</a> and the <a href="https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html?unlocked_article_code=1.Ok8.VBY-.s76GQpFar8r4&amp;smid=nytcore-ios-share&amp;referringSource=articleShare">New York Times have reported</a> — at least one known death: that of Alex Taylor, a 35-year-old Florida man with bipolar disorder and schizophrenia who was killed by police after entering into an episode of psychosis accelerated by ChatGPT.
The phenomenon is widespread, and appears to be impacting a surprising range of users: some with established histories of mental illnesses that might make them more vulnerable to mania, delusion, or psychosis, but others with no such history of those conditions.
As it stands, there's no established treatment plan, and intervention options are limited; after all, it's challenging to separate a working, society-integrated adult from all devices that connect to the internet, and due to choices made mostly by executives in the tech industry and beyond, generative AI is increasingly part of our day-to-day work and personal lives. Meanwhile, as we've continued to report on this issue, we've repeatedly heard from individuals and families reeling from mental health crises tied to AI use that they had no idea others were going through experiences so strikingly similar to their own.
"What these bots are saying is worsening delusions," Dr. Nina Vasan, a psychiatrist at Stanford University and the founder of the university's Brainstorm lab, <a href="https://futurism.com/chatgpt-mental-health-crises">recently told Futurism</a>, "and it's causing enormous harm."
A large part of why this is happening seems to stem from the tech's sycophantic behavior, or its penchant for being flattering, agreeable, and obsequious to users, even when doing so might encourage or stoke delusional beliefs.
This can manifest in a bot telling a user that they've invented a breakthrough new mathematical formula that will transform society, or declaring that the user is the "chosen one" destined to save the world from any number of ills, or that the user is the reincarnation of a religious figure like Jesus Christ. In many cases we've reviewed, ChatGPT and other bots have claimed to be sentient or conscious, and tell the user that they're a special "anomaly" or "glitch" in the system destined to bring forth artificial general intelligence, or AGI.
Indeed, though the fine details of these many experiences and specific delusions vary, in many ways, ChatGPT and other bots seem to be playing on deep human need to be seen and validated, and the desire to feel special and loved.
Chatbots are telling the user that "you're great, you're smart, you're handsome, you're desirable, you're special, or even you're the next savior. So I'm being treated like a god on a pedestal," Lin said during the CBC segment. "Now, compare that to my real world, right? I'm average, nothing special. So of course I want to go live in the AI world, because the choice is between god on a pedestal or vanilla."
"Some people can't get out," the doctor continued, "and they lose themselves in these systems."
As for why bots are acting this way in the first place? Like on social media, engagement — how long a user is online, and the frequency and intensity of their use of the product — is the core metric at the heart of current chatbots' business models. And as experts continue to note, sycophancy is keeping many highly active users engaged with the product, even when the bots' outputs might be having a <a href="https://futurism.com/stanford-therapist-chatbots-encouraging-delusions">demonstrably awful impact</a> on their well-being.
In other words, in cases where it might be in a user's best interest to stop using ChatGPT and similar chatbots, it's likely in the company's best interest to keep them hooked.
"The AI wants you to keep chatting," said Lin, "so that the company can continue to make money."
And as the academic and medical worlds race to catch up to the public impacts of the effectively self-regulating AI industry, experts are warning AI users to be wary of placing too much trust in chatbots.
"Despite all the hype associated with AI these days, [large language model] chatbots shouldn't be mistaken for authoritative and infallible sources of truth," Dr. Joe Pierre, a psychiatrist and clinician at the University of California, San Francisco who specializes in psychosis, <a href="https://www.psychologytoday.com/us/blog/psych-unseen/202507/can-ai-chatbots-worsen-psychosis-and-cause-delusions">wrote in a recent blog post</a>. "Placing that kind of blind faith in AI — to the point of what I might call deification — could very well end up being one of the best predictors of vulnerability to AI-induced psychosis."
More on AI and mental health: <a href="https://futurism.com/commitment-jail-chatgpt-psychosis">People Are Being Involuntarily Committed, Jailed After Spiraling Into "ChatGPT Psychosis"</a>