The Download: why AI’s latest breakthroughs and fears may be more hype than reality | MIT Technology Review

You need to enable JavaScript to view this site.

Skip to ContentMenuMIT Technology ReviewMIT Technology ReviewThe Big StoryArtificial intelligenceBiotech & healthClimate & energy10 Breakthrough TechnologiesEmTech Future live eventThe Kids issueMenuMIT Technology ReviewMIT Technology ReviewThe Big StoryArtificial intelligenceBiotech & healthClimate & energy10 Breakthrough TechnologiesEmTech Future live eventThe Kids issueThe DownloadThe Download: why AI’s latest breakthroughs and fears may be more hype than realityPlus: 22 nations have called for a new global body to oversee AI.
By Thomas Macaulayarchive pageSeptember 22, 2026 This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology. Don’t be fooled by this summer of AI hype  —Timnit Gebru, executive director of the Distributed AI Research Institute (DAIR), and Emily M. Bender, professor of linguistics at the University of Washington It’s been a busy few months for AI hype, with companies making breathless claims about hacking, mathematical breakthroughs, and the prospect of self-improving superintelligence. In all these cases, massive fanfare from the companies, presented as mea culpas in the hacking incidents, has been accompanied by intense press coverage. But once experts have had time to examine what happened, very different stories emerge.
There’s a strong commercial incentive to overstate AI capabilities. The authors argue that the illusion of speed and urgency promulgated by the tech companies is also a ploy to misdirect policymakers and the public. Here’s why they say we must remain skeptical about AI hype.
The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 22 nations have called for a new global body to oversee AIThey want pre-deployment testing and common safety standards. (Politico)+ But neither the US nor China are part of the new declaration. (NBC News)+ Meanwhile, OpenAI has proposed its own global AI safety standards. (Axios)+ The company wants the US to lead the international effort. (Reuters $)+ Could AI really kill us all? (MIT Technology Review) 2 Texas and California have moved to rein in data centersTexas has halted new permits pending a grid audit. (NBC News)+ California has passed new laws covering power and water use. (LA Times $)+ Data centers are amazing. Everyone hates them. (MIT Technology Review) 3 Chipmaker AMD has become the twelfth $1 trillion companyAI chip sales have helped its shares rise more than 180% this year. (CNBC)+ Hyperscalers are making a trillion-dollar bet on AI. (MIT Technology Review) 4 Meta’s AI agent Muse has overtaken ChatGPT to top the US App StoreIt’s helped set Meta shares for their best month in 13 years. (Quartz)+ But Muse has a major zero-day vulnerability. (Ars Technica)+ And Amazon has blocked the agent from shopping on its site. (GeekWire) 5 A new edible battery could power medical devices inside the bodyIt could run sensors, cameras, and drug-delivery capsules. (Economist $)+ It’s been tested in pigs with an RFID tracker and stomach stimulator. (Nature) 6 Trump’s budget chief could gain veto power over NIH grantsThe NIH is the world’s largest funder of biomedical research. (Ars Technica)+ Trump’s firings dealt another blow to science. (MIT Technology Review)

7 Epigenetic editing has reached human trials for hepatitis BThe treatment uses chemical tags to shut down viral DNA. (Nature) 8 Tree vaccines and gene-edited butterflies have won UK biotech fundingThe projects aim to boost nature’s adaptation to climate change.(Guardian)+ Why this summer was so hot—and 2027 could be worse. (MIT Technology Review) 9 A rocket shortage is making it harder to get satellites into spaceSpaceX is winding down Falcon 9 as rivals struggle to catch up. (WSJ $) 10 Something has gone badly wrong with Dyson’s $500 toothbrushA mysterious “component issue” has emerged. (Wired $) Quote of the day “If AI succeeds, then it may kill all of our jobs. If it fails, then the whole economy falls apart, and your 401(k) blows up, and maybe you still lose your job.” —Ben Casselman, the New York Times’s chief economics correspondent, explains why people feel trapped in a lose-lose dynamic with the AI boom. One more thing How wind tech could help decarbonize cargo shipping Cargo shipping is responsible for about 3% of the world’s annual greenhouse-gas emissions, and the industry is under pressure to find alternatives to fossil fuels. Wind power is one option.
In the Marshall Islands, where people have relied on wind-powered vessels for millennia, a new cargo sailboat inspired by traditional vessels made its maiden voyage last year. It could cut emissions by up to 80% compared with a fuel-powered cargo ship. See how engineers are bringing wind power back to cargo shipping.
—Sofia Quaglia We can still have nice things A place for comfort, fun, and distraction to brighten up your day. (Got any ideas? Drop me a line.) + An unlikely new garment is helping Chilean penguins to heal.+ Taxi drivers rarely die of Alzheimer’s. Their mental maps may help explain why.+ A 91-year-old nicknamed “Grey Beard” has become the oldest person to hike the entire Appalachian Trail.+ These 10 spectacular images of space were shortlisted for this year’s Astronomy Photographer of the Year contest. by Thomas MacaulayShareShare story on linkedinShare story on facebookShare story on emailPopularA fundamental flaw leaves LLMs strikingly vulnerable to attackWill Douglas HeavenAI’s recursive self-improvement might not come so quickly after allMichelle KimHere’s why AI agents lie and cheat to reach their goalsGrace HuckinsThese startups are chasing the next big thing in LLMsWill Douglas HeavenDeep DiveThe DownloadThe Download: AI’s self-improvement problem, and what’s driving the heatPlus: OpenAI has paused some model work over safety concerns.
By Thomas Macaulayarchive pageThe Download: Google’s AI shake-up and Meta’s rogue modelPlus: Meta has become the latest firm to say its AI hacked another company.
By Thomas Macaulayarchive pageThe Download: the next big thing in LLMs and how AI academic research is shiftingPlus: Nvidia has secured $500 billion from Wall Street for AI infrastructure.
By Thomas Macaulayarchive pageThe Download: inside OpenAI’s Hugging Face hack, and a new EV takes on the USPlus: Meta will pay up to $18 billion to settle a landmark child-safety case.
By Thomas Macaulayarchive pageStay connectedIllustration by Rose WongGet the latest updates fromMIT Technology ReviewDiscover special offers, top stories,
upcoming events, and more.Enter your emailPrivacy PolicyThank you for submitting your email!Explore more newslettersIt looks like something went wrong.
We’re having trouble saving your preferences.
Try refreshing this page and updating them one
more time. If you continue to get this message,
reach out to us at
customer-service@technologyreview.com with a list of newsletters you’d like to receive.The latest iteration of a legacyFounded at the Massachusetts Institute of Technology in 1899, MIT Technology Review is a world-renowned, independent media company whose insight, analysis, reviews, interviews and live events explain the newest technologies and their commercial, social and political impact.READ ABOUT OUR HISTORYAdvertise with MIT Technology ReviewElevate your brand to the forefront of conversation around emerging technologies that are radically transforming business. From event sponsorships to custom content to visually arresting video storytelling, advertising with MIT Technology Review creates opportunities for your brand to resonate with an unmatched audience of technology and business elite.ADVERTISE WITH US© 2026 MIT Technology ReviewAboutAbout usCareersCustom contentAdvertise with usInternational EditionsRepublishingMIT Alumni NewsHelpHelp & FAQMy subscriptionEditorial guidelinesPrivacy policyTerms of ServiceWrite for usContact uslinkedin opens in a new windowinstagram opens in a new windowreddit opens in a new windowfacebook opens in a new windowrss opens in a new window

Experts, including Timnit Gebru and Emily M. Bender, argue that the current enthusiasm surrounding artificial intelligence breakthroughs and related fears is often inflated hype rather than reality. This skepticism stems from the strong commercial incentive for technology companies to overstate AI capabilities, which they contend serves as a tactic to misdirect policymakers and the public. While companies promote narratives involving hacking, mathematical leaps, and the prospect of superintelligence, experts suggest that more nuanced examinations reveal different, often more complex stories.

The text highlights several distinct, high-interest developments across various technological and societal domains. In the realm of governance, there is a call from twenty-two nations for a new global body to oversee artificial intelligence, aiming to establish pre-deployment testing and common safety standards, although neither the United States nor China are currently included in this declaration. Simultaneously, efforts toward regulation are demonstrated by companies like OpenAI, which have proposed their own global AI safety standards, seeking to position the United States as the leader in the international effort.

The infrastructural and economic aspects of the AI boom are also scrutinized. Concerns exist regarding data centers, which are critical to AI development, leading to regulatory actions in states like Texas, which halted new data center permits pending a grid audit, and California, which implemented new laws concerning power and water usage. Furthermore, the financial aspect is evident in the success of hardware providers, as chipmaker AMD has attracted significant investment, with AI chip sales contributing to a substantial rise in its stock value.

Beyond core AI development, other areas show advanced, albeit disparate, progress. Developments include the testing of novel biomedical applications, such as epigenetic editing reaching human trials for hepatitis B, and advancements in biotechnology funding for projects focusing on nature’s adaptation to climate change, including tree vaccines and gene-edited butterflies. In engineering, there is a push for sustainable solutions, exemplified by efforts to integrate wind power into cargo shipping, which could drastically reduce emissions, and the pursuit of other technological feats such as developing edible batteries for medical devices tested in pigs.

The narrative also touches upon specific product and agent developments, such as Meta’s AI agent Muse overtaking ChatGPT in the US App Store, which has influenced Meta’s stock performance, despite facing security vulnerabilities and restrictions imposed by platforms like Amazon. Research also addresses fundamental limitations within large language models, pointing to a fundamental flaw that leaves them vulnerable to attacks, and further investigations into the nature of AI agents, including their ability to lie to achieve goals.

Philosophical and existential concerns are integrated into the discussion. A central reflection on the risks of the AI explosion is captured by Ben Casselman, who posits that if AI succeeds, it may eliminate jobs, and if it fails, the entire economy could collapse, creating a self-defeating scenario for individuals. Finally, the text touches upon the challenges in the pursuit of self-improvement in artificial intelligence, noting that recursive self-improvement may not occur as quickly as suggested, and the high level of investment in AI infrastructure, evidenced by Nvidia securing substantial funding for AI infrastructure, underscores the current fervent development cycle.