The Cybersecurity Information Sharing Act [1]expired on Wednesday when the federal government shut down. The law had provided legal protections since 2015 for organizations to share cyber threat intelligence with federal agencies. Without these protections, private sector companies that control most U.S. critical infrastructure face potential legal risks when sharing information about threats. Sen. Gary Peters called the lapse "an open invitation to cybercriminals and hostile actors to attack our economy and our critical infrastructure."

The intelligence sharing enabled by CISA 2015 helped expose Chinese campaigns including Volt Typhoon in 2023 and Salt Typhoon last year. Several cybersecurity firms pledged to continue sharing threat data despite the law's expiration. Halcyon and CrowdStrike confirmed they would maintain information sharing. Palo Alto Networks said it remained committed to public-private partnerships but did not specify whether it would continue sharing threat data. Multiple bipartisan reauthorization efforts failed before the shutdown. The House Homeland Security Committee had approved a 10-year extension last month.



[1] https://www.politico.com/news/2025/10/03/cyber-law-cisa-2015-shutdown-00592501

Jeff Bezos told an audience on Friday that gigawatt-scale data centers will be built in space [1]within the next ten to twenty years . The Amazon founder said these orbital facilities would eventually outperform their terrestrial counterparts because space offers uninterrupted solar power around the clock.

Bezos was speaking in a fireside chat with Ferrari and Stellantis Chairman John Elkann. He said the giant training clusters needed for AI would be better built in space because there are no clouds, rain or weather to interrupt power generation. Bezos predicted that space-based data centers would beat the cost of Earth-based ones within a couple of decades. He described the shift as part of a broader pattern that has already occurred with weather satellites and communication satellites. The next steps would be data centers and then other kinds of manufacturing.



[1] https://www.reuters.com/business/energy/data-centres-space-jeff-bezos-thinks-its-possible-2025-10-03/

Air pollution increases the likelihood of people [1]becoming frail in middle and old age , according to an international review of studies. The Guardian:

> The review team found 10 studies that looked at outdoor air pollution and frailty. The people studied came from 11 countries including China, the UK, Sweden, South Africa and Mexico. Two of the studies showed that men were more vulnerable than woman, with a stronger association between particle pollution and frailty. The risk of frailty increased with outdoor particle pollution. For the UK, this could mean about 10-20% of frailty cases are attributable to air pollution.

>

> Exposure to secondhand smoking was the environmental factor that presented the greatest risk of frailty. The risk of frailty was increased by about 60% for people who breathed other people's smoke at home. Using solid fuels for cooking or home heating also carried an extra risk of frailty. This was about half the risk of living with a smoker, based on studies from six countries.



[1] https://www.theguardian.com/environment/2025/oct/03/frailty-in-ageing-populations-worsened-by-air-pollution-global-review-finds

Pew Research:

> Public awareness of legal sports betting has grown in recent years -- and [1]so has the perception that it is a bad thing for society and sports , according to a new Pew Research Center survey. Today, 43% of U.S. adults say the fact that sports betting is now legal in much of the country is a bad thing for society. That's up from 34% in 2022. And 40% of adults now say it's a bad thing for sports, up from 33%.

>

> Despite these increasingly critical views of legal sports betting, many Americans continue to say it has neither a bad nor good impact on society and on sports. Fewer than one-in-five see positive impacts. Meanwhile, the share of Americans who have bet money on sports in the past year has not changed much since 2022.

>

> Today, 22% of adults say they've personally bet money on sports in the past year. That's a slight uptick from 19% three years ago. This figure includes betting in any of three ways:

> 1. With friends or family, such as in a private betting pool, fantasy league or casual bet

> 2. Online with a betting app, sportsbook or casino

> 3. In person at a casino, racetrack or betting kiosk

Further reading : [2]Filipinos Are Addicted to Online Gambling. So Is Their Government .



[1] https://www.pewresearch.org/short-reads/2025/10/02/americans-increasingly-see-legal-sports-betting-as-a-bad-thing-for-society-and-sports/

[2] https://news.slashdot.org/story/25/10/01/1755259/filipinos-are-addicted-to-online-gambling-so-is-their-government

James Marriott, [1]writing in a column :

> The world of print is orderly, logical and rational. In books, knowledge is classified, comprehended, connected and put in its place. Books make arguments, propose theses, develop ideas. "To engage with the written word," the media theorist Neil Postman wrote, "means to follow a line of thought, which requires considerable powers of classifying, inference-making and reasoning."

>

> As Postman pointed out, it is no accident, that the growth of print culture in the eighteenth century was associated with the growing prestige of reason, hostility to superstition, the birth of capitalism, and the rapid development of science. Other historians have linked the eighteenth century explosion of literacy to the Enlightenment, the birth of human rights, the arrival of democracy and even the beginnings of the industrial revolution. The world as we know it was forged in the reading revolution.

>

> Now, we are living through the counter-revolution. More than three hundred years after the reading revolution ushered in a new era of human knowledge, books are dying. Numerous studies show that reading is in free-fall. Even the most pessimistic twentieth-century critics of the screen-age would have struggled to predict the scale of the present crisis. In America, reading for pleasure has fallen by forty per cent in the last twenty years. In the UK, more than a third of adults say they have given up reading. The National Literacy Trust reports "shocking and dispiriting" falls in children's reading, which is now at its lowest level on record. The publishing industry is in crisis: as the author Alexander Larman writes, "books that once would have sold in the tens, even hundreds, of thousands are now lucky to sell in the mid-four figures."

>

> [...] What happened was the smartphone, which was widely adopted in developed countries in the mid-2010s. Those years will be remembered as a watershed in human history. Never before has there been a technology like the smartphone. Where previous entertainment technologies like cinema or television were intended to capture their audience's attention for a period, the smartphone demands your entire life. Phones are designed to be hyper-addictive, hooking users on a diet of pointless notifications, inane short-form videos and social media rage bait.



[1] https://jmarriott.substack.com/p/the-dawn-of-the-post-literate-society-aa1

Social media usage peaked in 2022 and has been [1]on a steady decline since . An analysis of 250,000 adults across more than 50 countries by the digital audience insights company GWI found that adults aged 16 and older spent an average of two hours and 20 minutes per day on social platforms at the end of 2024. That figure is down almost 10% from 2022. The decline is most pronounced among teenagers and people in their twenties.

Usage has traced a smooth curve upward and then downward over the past decade. This is not simply the unwinding of increased screen time during pandemic lockdowns. The data also captured a shift in how people use these platforms. The share of people who report using social media to stay in touch with friends, express themselves or meet new people has fallen by more than a quarter since 2014.

Opening the apps reflexively to fill spare time has risen. North America is an exception to the global trend. Social media consumption there continues to climb. By 2024 it reached levels 15% higher than Europe. Meta and OpenAI recently announced new social platforms that will be filled with AI-generated short-form videos.



[1] https://www.ft.com/content/a0724dd9-0346-4df3-80f5-d6572c93a863

Ha Dang, a self-taught accountant from Scunthorpe who trained via YouTube, [1]won the inaugural Microsoft Excel UK Championships on September 30. The victory earned him a spot at the Microsoft Excel World Championships in Las Vegas, a three-day tournament inside a 30,000-square-foot esports arena where players compete for $5,000 and are broadcast on ESPN.

Thirty competitors sat shoulder to shoulder through three gruelling rounds of spreadsheet challenges. Each round featured a custom case with seven levels of increasing difficulty. The second round case, Right Royal Battle Part II, took 80 drafts to perfect. Players calculated troop sizes from emoji battalions and army movements across fourteenth-century France. Hadyn Wiseman, who once held the Guinness World Record for most backflips in a minute, placed fourth. Lara Holding-Jones finished thirteenth. Jaq Kennedy founded the UK chapter last year. National chapters have since formed in Germany, Brazil, and Chile.



[1] https://www.huckmag.com/article/microsoft-excel-championships-world-uk-ha-deng-vegas-40-anniversary-obsessives

Snapchat will [1]start charging users who exceed 5GB of saved Memories , with paid plans starting at $1.99/month for 100GB. "If your memories exceed this limit, you'll need to subscribe to one of its new Memories Storage plans," reports TechCrunch. From the report:

> The company told TechCrunch in an email that the introductory storage plan offers up to 100GB of storage for $1.99 per month. Snapchat+ users will get up to 250GB of storage as part of their $3.99 monthly subscription, while Snapchat Platinum users will get 5TB as part of their $15.99 monthly subscription. Snapchat explains that when it first launched Memories, it didn't expect it to grow to what it has today, as users have saved more than 1 trillion Memories on the platform.

>

> Snapchat will provide 12 months of temporary Memories storage for any Memories that exceed the 5GB storage limit. The company notes that users can download Memories directly to their devices. If you're over the limit, but don't sign up for a plan, your oldest Snaps will be saved, while the most recent ones that are over the storage limit will be deleted. Snapchat says the change won't affect most users, as the vast majority have under 5GB of Memories. It will mainly impact those with "thousands of Snaps," the company notes.

"It's never easy to transition from receiving a service for free to paying for it, but we hope the value we provide with Memories is worth the cost," Snapchat wrote in a blog post . "These changes will allow us to continue to invest in making Memories better for our entire community."



[1] https://techcrunch.com/2025/09/29/snapchat-caps-free-memory-storage-launches-paid-storage-plans/

Longtime Slashdot reader [1]theodp writes:

> [2]Big Tech Told Kids to Code. The Jobs Didn't Follow , a New York Times podcast episode discussing how the [3]promise of a six-figure salary for those who study computer science is turning out to be an empty one for recent grads in the age of AI, [4]drew the ire of the co-founders of nonprofit Code.org , which -- ironically -- is [5]pivoting to AI itself with the encouragement of, [6]and millions from , its [7]tech-giant backers .

>

> In a [8]LinkedIn post , Code.org CEO and co-founder Hadi Partovi said the paper and its Monday episode of "The Daily" podcast were cherrypicking anecdotes "to stoke populist fears about tech corporations and AI." He also took to X, [9]tweeting : "Today the NYTimes (falsely) claimed [10]CS majors can't find work . The data tells the opposite story: CS grads have the highest median wage and the fifth-lowest underemployment across all majors. [...] Journalism is broken. Do better NYTimes." To which Code.org co-founder Ali Partovi (Hadi's twin), [11]replied : "I agree 100%. That NYTimes Daily piece was deplorable -- an embarrassment for journalism."



[1] https://slashdot.org/~theodp

[2] https://www.nytimes.com/2025/09/29/podcasts/the-daily/big-tech-told-kids-to-code-the-jobs-didnt-follow.html?unlocked_article_code=1.p08.40RL.5RkndIfsAQEe&smid=url-share

[3] https://web.archive.org/web/20150321035824/https://www.seattletimes.com/opinion/guest-how-to-prep-more-students-for-computer-science-careers/

[4] https://www.geekwire.com/2025/code-org-ceo-rips-ny-times-for-stoking-populist-fears-over-computer-science-jobs-and-ai/

[5] https://developers.slashdot.org/story/25/08/10/0110212/hour-of-code-announces-its-now-evolving-into-hour-of-ai

[6] https://blog.google/outreach-initiatives/education/ai-education-efforts/

[7] https://code.org/en-US/about/supporters

[8] https://www.linkedin.com/feed/update/urn:li:activity:7378663638309580800/

[9] https://x.com/hadip/status/1972899445743473085

[10] https://developers.slashdot.org/story/25/09/29/029201/professor-warns-cs-graduates-are-struggling-to-find-jobs

[11] https://x.com/apartovi/status/1972900594869588434

An anonymous reader quotes a report from MIT Technology Review:

> A team at Microsoft says it used artificial intelligence to [1]discover a "zero day" vulnerability in the biosecurity systems used to prevent the misuse of DNA. These screening systems are designed to stop people from purchasing genetic sequences that could be used to create deadly toxins or pathogens. But now researchers led by Microsoft's chief scientist, Eric Horvitz, says they have figured out how to bypass the protections in a way previously unknown to defenders.The team [2]described its work today in the journalScience .

>

> Horvitz and his team focused on generative AI algorithms that propose new protein shapes. These types of programs are already fueling the hunt for new drugs at well-funded startups like Generate Biomedicines and Isomorphic Labs, a spinout of Google. The problem is that such systems are potentially "dual use." They can use their training sets to generate both beneficial molecules and harmful ones. Microsoft says it began a "red-teaming" test of AI's dual-use potential in 2023 in order to determine whether "adversarial AI protein design" could help bioterrorists manufacture harmful proteins.

>

> The safeguard that Microsoft attacked is what's known as biosecurity screening software. To manufacture a protein, researchers typically need to order a corresponding DNA sequence from a commercial vendor, which they can then install in a cell. Those vendors use screening software to compare incoming orders with known toxins or pathogens. A close match will set off an alert. To design its attack, Microsoft used several generative protein models (including its own, called EvoDiff) to redesign toxins -- changing their structure in a way that let them slip past screening software but was predicted to keep their deadly function intact.

"This finding, combined with rapid advances in AI-enabled biological modeling, demonstrates the clear and urgent need for enhanced nucleic acid synthesis screening procedures coupled with a reliable enforcement and verification mechanism," says Dean Ball, a fellow at the Foundation for American Innovation, a think tank in San Francisco.



[1] https://www.technologyreview.com/2025/10/02/1124767/microsoft-says-ai-can-create-zero-day-threats-in-biology/

[2] https://www.science.org/content/article/made-order-bioweapon-ai-designed-toxins-slip-through-safety-checks-used-companies

[1]alternative_right shares a report from the Smoking Gun:

> Minutes after vandalizing 17 cars in a Missouri college parking lot, a 19-year-old sophomore had a [2]lengthy ChatGPT conversation during which he confessed to the crime , asked about the possibility of getting caught, and wondered, "is there any way they could know it was me," according to a police probable cause [3]statement . Ryan Schaefer was arrested yesterday and charged with felony property damage for a rampage early Sunday at a Missouri State University parking lot. Investigators allege that Schaefer shattered car windows, ripped off side mirrors, dented hoods, and broke windshield wipers during the 3 AM spree.

>

> When confronted with surveillance footage and other evidence, Schaefer said that he could see the resemblance between the suspect and himself. At that point, Schaefer reportedly consented to a search of his iPhone. A subsequent review of the device revealed location data placing Schaefer "at or near the scene of the crime," as well as a "troubling dialogue exchange this defendant seems to have had with artificial intelligence software installed on his phone," prosecutors reported.

The incriminating ChatGPT conversation can be found [4]here .



[1] https://slashdot.org/~alternative_right

[2] https://www.thesmokinggun.com/documents/bizarre/chatgpt-confession-196015

[3] http://www.thesmokinggun.com/file/chatgpt-confession

[4] https://www.thesmokinggun.com/file/chatgpt-confession?page=1

[1]Bruce66423 shares a report from The Guardian:

> Those diagnosed as small children [2]typically have distinct genetic profile from those diagnosed later , [finds an international study based on genetic data from more than 45,000 autistic people in Europe and the U.S]. So, there's more than one condition out there that's being diagnosed as "autism." This, of course, messes with the [3]debate about causes ; one version of autism may be caused by something for which the evidence is very weak overall.

"The term 'autism' likely describes multiple conditions," said Dr Varun Warrier, from Cambridge's department of psychiatry, senior author of the research. "For the first time, we have found that earlier and later diagnosed autism have different underlying biological and developmental profiles."

"It is a gradient," added Warrier. "There are also many other factors that contribute to age of diagnosis, so the moment you go from averages to anything that is applicable to an individual, it's false equivalency."

The analysis has been [4]published in the journal Nature .



[1] https://slashdot.org/~Bruce66423

[2] https://www.theguardian.com/society/2025/oct/01/autism-should-not-be-seen-as-single-condition-with-one-cause-say-scientists

[3] https://en.wikipedia.org/wiki/Controversies_in_autism

[4] https://www.nature.com/articles/s41586-025-09542-6

Ford's push for a four-day in-office workweek hit turbulence when someone hijacked meeting room screens to [1]display an anti-RTO protest image targeting CEO Jim Farley . The company quickly removed it and is investigating. The Detroit Free Press reports:

> According to photos employees took of the image, which were [2]posted on social media and sent to the Detroit Free Press, it contained an image of CEO Jim Farley along with a big red circle with a slash through it over his face and the words "(Expletive) RTO."

>

> "We're aware of an inappropriate use of Ford's IT technology and we're investigating it," Dave Tovar, Ford spokesman, told the Detroit Free Press. Tovar said the image was up for "a short amount of time" and Ford was able to quickly remove it. He said the company is investigating whether the image appeared only in Dearborn offices or globally.

>

> Farley [3]mandated that employees return to the office four days a week earlier this year and it has been in place since Sept. 1, with no fallout such as people quitting over it, Tovar said. Therefore, Tovar said, "I wouldn't be able to speculate on it, as to why someone would do this."



[1] https://www.yahoo.com/news/articles/ford-system-tampered-display-vulgar-183054042.html

[2] https://x.com/caseysfloyd/status/1973849814136992042

[3] https://www.reuters.com/business/autos-transportation/ford-calls-majority-workforce-into-office-four-days-week-2025-06-25/

Colorado unveiled an autonomous crash-protection truck [1]designed to absorb highway work zone collisions , removing human drivers from one of the most dangerous road maintenance roles. The Denver Gazette reports:

> At a press event in Falcon on Monday, the Colorado Department of Transportation demonstrated an autonomous truck-mounted attenuator -- a driverless crash-protection vehicle designed to absorb impacts in roadside work areas. These trucks are already in the state's fleet, but previously required a worker in the cab, leaving them exposed during crashes. "These vehicles are designed to get hit so people don't have to," said Kay Kelly, CDOT's chief of innovative mobility. "We want to remove the human from that truck whenever possible."

>

> [...] Colorado pioneered this technology in 2017, becoming the first state to test it. Developed by San Diego-based defense contractor Kratos Defense, the company started on autonomous follower systems in 2012 and installed its initial version in Colorado in 2013, according to Maynard Factor, Kratos' vice president of business development. The system uses sensors, actuators, and video links, allowing the unmanned crash truck to trail directly behind a paint-striping or maintenance vehicle. The lead vehicle records its route and transmits navigation data to the follower. Both are equipped with global positioning system and communication tools, with the lead driver monitoring via video feed.

>

> "It's a convoy system," Factor said. "The leader does the work, while the driverless vehicle mirrors its path. This protects the operator with the crash truck instead of sitting inside it." Sensors detect objects darting into the lane, triggering an automatic stop. Operators can pause or override via a user interface featuring forward-facing cameras on the autonomous truck and rear-facing ones on the work vehicle. The trucks, existing state assets, are driven to sites and then switched to autonomous mode. Outfitting each with the technology costs about $1 million, excluding the base vehicle and crash absorber.



[1] https://www.denvergazette.com/2025/09/30/self-driving-cdot-crash-truck-protects-highway-workers/

An anonymous reader quotes a report from The Register:

> New Zealand's Institute of IT Professionals has discovered it is insolvent and [1]advised members it has no alternative but to enter liquidation . The Institute (ITP) wrote to members on Thursday and [2]posted a document titled "Important Update on ITP's Future" that reveals it has "reached a point where the organization cannot continue. After a full review of our finances, the Board has confirmed that ITP is insolvent."

>

> Insolvency seems to have come as something of a surprise. "These debts are historic. They go back over many years. While some of the issues were worked on in more recent times, the full scale of the problem only became visible during the leadership change in 2025," the Update states. "Once the Board understood the full picture, it was clear that there was no responsible way forward other than liquidation." [...]

>

> ITP's constitution requires its members to formally resolve to wind up the organization, so as one of its final acts the group has called a Special General Meeting (SGM) for 23 October 2025 to confirm liquidation and appoint a liquidator. This situation impacts more than ITP's ~10,000 members, because the organization offers assessment services that assess whether IT professionals' skills and qualifications make them eligible to move to New Zealand for work. ITP also certifies IT degrees at New Zealand universities, and oversees the NZ Cloud Computing Code of Practice. ITP also conducted educational and advocacy activities aimed at growing New Zealand's tech workforce.



[1] https://slashdot.org/NewZealand'sInstituteofITProfessionals

[2] https://itp.nz/closure

"Your AMD chips may have Intel Inside soon," writes longtime Slashdot reader [1]DesScorp . "Discussions are underway between the two companies to [2]move an undisclosed amount of AMD's chip business to Intel foundries . (AMD currently does their production through TSMC.) The talks come hot on the heels of a flurry of [3]other Intel investments ." Tom's Hardware reports:

> In the past several weeks, Intel has seen a flurry of activity and investments. The United States [4]announced a 9.9% ownership stake in Intel, while Softbank [5]bought $2 billion worth of shares. Alongside Nvidia, Intel announced new x86 chips using Nvidia graphics technology, with the graphics giant also purchasing $5 billion in Intel shares. There have also been reports that Intel and Apple have been [6]exploring ways to work together.

The article notes that there is a trade/political dimension to an AMD-Intel deal as well:

> It makes sense for Intel's former rivals -- especially American companies -- to consider coming to the table. The White House is pushing for 50% of chips bound for America to be built domestically, and tariffs on chips aren't off the table. Additionally, doing business with Intel could make the US government, Intel's largest shareholder, happy, which can be good for business. AMD faced export restrictions on its GPUs earlier this year as the US attempted to throttle China's AI business.



[1] https://slashdot.org/~DesScorp

[2] https://www.tomshardware.com/pc-components/cpus/amd-in-early-talks-to-make-chips-at-intel-foundry-report-says

[3] https://slashdot.org/story/25/09/18/1226210/nvidia-to-invest-5-billion-in-intel

[4] https://news.slashdot.org/story/25/08/22/1849210/intel-has-agreed-to-a-deal-for-us-to-take-10-equity-stake-trump-says

[5] https://slashdot.org/story/25/08/18/2337225/intel-is-getting-a-2-billion-investment-from-softbank

[6] https://apple.slashdot.org/story/25/09/24/2141256/intel-approaches-apple-for-potential-investment-amid-struggles

Two Amazon Prime Air drones [1]collided with a crane in Tolleson, Arizona near 96th Avenue and Roosevelt Street. Amazon confirmed the incident and is working with authorities to determine what happened, though no injuries have been reported. CNBC reports:

> The incident occurred on Wednesday around 1 p.m. EST in Tolleson, Arizona, a city west of Phoenix. Two MK30 drones crashed into the boom of a stationary construction crane that was in a commercial area just a few miles away from an Amazon warehouse. One person was evaluated on the scene for possible smoke inhalation, said Sergeant Erik Mendez of the Tolleson Police Department.

>

> Both drones sustained "substantial" damage from the collision on Wednesday, which occurred when the aircraft were mid-route, according to preliminary FAA crash reports. The Federal Aviation Administration and National Transportation Safety Board are investigating the incident. The drones were believed to be flying northeast back-to-back when they collided with the crane that was being used for roof work on a distribution facility, Tolleson police said in a release. The drones landed in the backyard of a nearby building, according to the release.



[1] https://www.abc15.com/news/region-west-valley/tolleson/two-amazon-delivery-drones-crash-into-crane-in-commercial-area-of-tolleson

An anonymous reader quotes a report from TorrentFreak:

> The operator of a popular pirate sports streaming site in Argentina has gone from spending time in jail with murderers to [1]landing a new high-profile job a month later . Alejo "Shishi" Warles, the 25-year-old operator of Al Angulo TV, was arrested on August 20 in a LaLiga-backed crackdown. After his release on bail, he was hired by professional esports team 9z Globant, a partnership involving Argentine tech unicorn Globant. [...] The team is the result of a partnership between 9z Team and Argentinian tech unicorn Globant. Somewhat ironically, Globant previously worked with LaLiga to monitor the live-streaming user experience. Warles [2]welcomed himself to 9z Globant via the team's social media account, referring to himself as an idol, genius, and GOAT.

>

> Lucia Quinteros, the main social media manager at the esports team, [3]informed Entre Rios that after considering their new hire's history, they believe that he can add value to the team. "We hired Alejo, not the person who set up that project (Al Angulo TV). Of course, we evaluated what happened, but we believe that, from now on, Alejo can pursue a different career path," Quinteros said. According to Warles himself, he was hired because he's the best. Like many of his comments, this bravado should not be taken too seriously, but nevertheless sits in stark contrast to the typical pirate site operator facing criminal charges.



[1] https://torrentfreak.com/sports-piracy-operator-goes-from-jail-to-getting-hired-by-a-tech-unicorn-in-a-month/

[2] https://x.com/9zTeam/status/1968821005855359467

[3] https://www.elentrerios.com/actualidad/el-ldquoautodidactardquo-que-cre-una-web-pirata-desde-entre-ros-pas-de-estar-preso-a-ser-contratado-por-un-unicornio-argentino.htm

The Hunga Tonga-Hunga Ha'apai volcano erupted on January 15, 2022. The pyroclastic flow severed both of Tonga's underwater internet cables. The eruption cut sixty-five miles from the domestic cable and fifty-five miles from the international link to Fiji. Tonga lost all internet access. The cables sit on the ocean floor and carry 95% of the world's international internet traffic.

The Guardian has [1]a long read on what happened in the aftermath . A.T.M.s (cash machines) stopped working because banks could not verify account balances. Businesses could not file export paperwork. Foreign remittances made up 44% of the country's G.D.P. The government found old satellite phones. Three or four days later, officials restored a hundred and twenty megabytes per second of bandwidth for essential work. A month after the eruption, SpaceX donated fifty Starlink terminals. SubCom's repair ship Reliance took five weeks to restore the international cable. Vava'u did not get broadband back until August, 2023. Another earthquake in the summer of 2024 severed the domestic cable again.



[1] https://www.theguardian.com/news/2025/sep/30/tonga-pacific-island-internet-underwater-cables-volcanic-eruption

AI has [1]run out of training data , according to Neema Raphael, Goldman Sachs' chief data officer and head of data engineering. "We've already run out of data," Raphael said on the bank's podcast. He said this shortage is already shaping how developers build new AI systems. China's DeepSeek may have kept costs down by training on outputs from existing models instead of fresh data. The web has been tapped out.

Developers have been using synthetic data -- machine-generated material that offers unlimited supply but carries quality risks. Raphael said he doesn't think the lack of fresh data will be a massive constraint. "From an enterprise perspective, I think there's still a lot of juice I'd say to be squeezed in that," he said. Proprietary datasets held by corporations could make AI tools far more valuable. The challenge is "understanding the data, understanding the business context of the data, and then being able to normalize it."



[1] https://www.businessinsider.com/ai-training-data-shortage-slop-goldman-sachs-2025-10