DJ Henderson – Page 3

Rocket Report: China launches with no advance warning; Europe’s drone ship

rocket report, Space / DJ Henderson / October 17, 2025

Starlink, Kuiper, and the US military all saw additions to their mega-constellations this week.

SpaceX’s Starship descends toward the Indian Ocean at the conclusion of Flight 11. Credit: SpaceX

Welcome to Edition 8.15 of the Rocket Report! This year has been, at best, one of mixed results for SpaceX’s Starship program. There have been important steps forward, including the successful reuse of the rocket’s massive Super Heavy booster. Clearly, SpaceX is getting really good at launching and recovering the 33-engine booster stage. But Starship itself, part spacecraft and part upper stage, hasn’t fared as well—at least it hadn’t until the last couple of months. After four Starships were destroyed in flight and on the ground in the first half of 2025, the last two missions ended with pinpoint splashdowns in the Indian Ocean. The most recent mission this week was arguably the most successful yet for Starship, which returned to Earth with little damage, suggesting SpaceX’s improvements to the heat shield are working.

As always, we welcome reader submissions. If you don’t want to miss an issue, please subscribe using the box below (the form will not appear on AMP-enabled versions of the site). Each report will include information on small-, medium-, and heavy-lift rockets, as well as a quick look ahead at the next three launches on the calendar.

SpaceX vet will fly with Blue Origin. Hans Koenigsmann is one of SpaceX’s earliest, longest-tenured, and most-revered employees. He worked at Elon Musk’s space company for nearly two decades, rising to the role of vice president for mission assurance and safety before leaving SpaceX in 2021. He led the investigations into every Falcon rocket failure, mentored young engineers, and became a public face for SpaceX through numerous presentations and press conferences. And now he has announced he is going to space on a future suborbital flight on Blue Origin’s New Shepard vehicle, Ars reports.

Due diligence … Koenigsmann will fly to space alongside his friend Michaela “Michi” Benthaus as early as next month. She’s notable in her own right—a mountain biking accident in 2018 left her with a spinal cord injury, but she did not let this derail her from her dream. She will become the first wheelchair user to fly in space. Koenigsmann said one of his main concerns with the flight was safety, but meeting with Blue Origin engineers gave him confidence to climb aboard New Shepard. “When we met them, I asked a lot of technical questions on the safety side, and I feel like they answered the majority of them thoughtfully and correctly.” So, what’s it like for a long-time SpaceXer to work with a former competitor, Blue Origin? Read Eric Berger’s interview with Koenigsmann to learn more.

The easiest way to keep up with Eric Berger’s and Stephen Clark’s reporting on all things space is to sign up for our newsletter. We’ll collect their stories and deliver them straight to your inbox.

Sign Me Up!

Europe’s drone ship. The European Space Agency (ESA) has awarded a contract for the design of a reusable rocket stage recovery vessel to the Italian aerospace and defense systems company Ingegneria Dei Sistemi (IDS), European Spaceflight reports. The project is part of a broader contract awarded to the Italian rocket builder Avio for the development of a reusable rocket upper stage, which Ars reported on last month. The contract covers preliminary design work for the launch system and the ground system, and could be applied to a reusable evolution of Avio’s Vega family of rockets.

Looks familiar … On Wednesday, IDS announced that it had been awarded the contract to design the project’s recovery vessel, which falls under the systems ground segment. The company has subcontracted Italian naval systems consultancy Cetena and Norwegian shipbuilder Vard to assist with the project. An artist’s illustration of the vessel gives it a familiar look. It appears similar to the recovery ships that SpaceX used to attempt recovery of the Falcon 9 rocket’s payload fairings, with giant nets to catch the hardware falling from space under parachute. (submitted by EllPeaTea)

JAXA looks abroad. The Japanese space agency JAXA has selected Rocket Lab to launch a set of technology demonstration satellites on Electron rockets after continued delays with a Japanese launch vehicle, Space News reports. The agreement covers two launches from New Zealand, the first in December with JAXA’s 242-pound (110-kilogram) Rapid Innovative Payload Demonstration Satellite-4 (RAISE-4) technology demonstration satellite, and the second in early 2026 with a batch of eight smaller satellites for educational, ocean monitoring, and other demonstrations.

No more waiting … These satellites were supposed to launch on Japan’s solid-fueled Epsilon S rocket, but JAXA looked to another launch provider after lengthy delays with the Epsilon program. Epsilon S is an upgraded version of Japan’s Epsilon rocket, which has flown six times. The first flight of Epsilon S was originally expected in 2023, but back-to-back ground test failures of the vehicle’s second stage solid rocket motor have effectively grounded the rocket. Japanese officials are considering ditching the upgraded second stage design and going back to the original Epsilon configuration, but a launch is still at least a year away.

An update on a German launch startup. German rocket builder HyImpulse announced Thursday that it had secured $53 million (45 million euros) in new funding to continue developing its SL-1 rocket, European Spaceflight reports. HyImpulse said it will use the new capital to “drive forward the development and commercialization of the SL1 orbital rocket and expand its production capacities.” HyImpulse is one of a handful of serious European launch startups, having raised more than $86 million (74 million euros) since its foundation in 2018.

Still years away … The SL1 rocket will consist of three stages with hybrid propulsion, capable of delivering up to more than 1,300 pounds (600 kilograms) of payload to low-Earth orbit. The first flight of HyImpulse’s orbital rocket is scheduled for 2027. SL1 builds on the company’s SR75 suborbital rocket, which made its first test flight from Australia in 2024.

iRocket touts rapid build. Innovative Rocket Technologies Inc. (iRocket) reports a successful flight test of the company’s 2.75-inch (70-millimeter) diameter IRX-100 version of the Hydra 70 rocket system from a launch tube under its own power to exercise a range of motor and missile properties, Aviation Week & Space Technology Reports. The IRX-100 is iRocket’s version of the Hydra 70 short-range unguided missile primarily used on military helicopters. Asad Malik, iRocket’s CEO, wrote in a post on LinkedIn that the company designed and launched the rocket in just 30 days. “Speed, precision, and innovation are what define our team,” Malik wrote.

Pathfinder … The IRX-100 rocket launched from a desert location in California and reached an altitude of more than 12,000 feet, according to iRocket. We’ve reported on iRocket in several recent editions of the Rocket Report. In July, the company announced it was going public in a deal with a Special Purpose Acquisition Company founded by former Commerce Secretary Wilbur Ross. But the SPAC and iRocket itself appear to have little money. Company officials hope the IRX-100 might offer a short-term source of revenue through military sales. iRocket’s longer-term goals include the development of a reusable orbital-class rocket, named Shockwave.

SpaceX launches for Kuiper. After more than a week of launch delays, SpaceX launched a Falcon 9 rocket from Cape Canaveral Space Force Station, Florida, with two dozen of Amazon’s Project Kuiper broadband Internet satellites onboard Monday night, Spaceflight Now reports. The mission, dubbed Kuiper Falcon 03 or KF-03, faced several days of launch delays due to poor weather both at the Cape as well as offshore. This was the third and final Kuiper launch currently booked on SpaceX’s Falcon 9 rocket, and the sixth launch of operational Kuiper satellites overall. Amazon now has 153 of its planned 3,232 Kuiper satellites in orbit.

SDA, too … Two days later, SpaceX launched a different Falcon 9 rocket from Vandenberg Space Force Base, California, to add 21 satellites to the Space Development Agency’s burgeoning low-Earth orbit constellation, Spaceflight Now reports. These satellites were built by Lockheed Martin, and they will join a batch of 21 similar spacecraft manufactured by York Space Systems launched last month. The satellites form the foundation for the Pentagon’s proliferated missile tracking and data relay network.

China launches another mysterious satellite. China conducted an orbital launch Monday with no apparent advance indication, successfully sending the Shiyan-31 remote sensing test satellite into orbit, Space News reports. The mission lifted off aboard a Long March 2D rocket from Jiuquan Satellite Launch Center in northwestern China. The Long March 2D can deliver up to 3.5 metric tons (7,700 pounds) of payload to low-Earth orbit. Shiyan-31 is believed to have an optical surveillance mission, and US tracking data indicated it was flying in an orbit about 300 miles (500 kilometers) above the Earth.

Surprise! … What was unusual about this launch was the fact that China did not publicize it in advance. Like most spacefaring nations, China typically issues airspace and maritime warning notices for airplanes and ships to steer clear of downrange zones where rocket debris may fall. No such warnings were released for this launch.

Starship flirts with perfection. SpaceX closed a troubled but instructive chapter in its Starship rocket program Monday with a near-perfect test flight that carried the stainless steel spacecraft halfway around the world from South Texas to the Indian Ocean, Ars reports. This was the 11th full-scale test flight of the Super Heavy booster and Starship upper stage, and it was arguably the most successful Starship test flight to date. It comes after a rough start to the year with a series of Starship failures and explosions that set the program back by at least six months.

Close to pristine … This time, Starship came back through the atmosphere with little sign of visible damage. The previous test flight in August also nailed its splashdown in the Indian Ocean, but it came down with a banged-up heat shield. This was the final flight of the second generation of Starship, called Starship V2. SpaceX plans to debut the larger, more powerful Starship V3 configuration in early 2026. If all goes well, SpaceX could be in position to attempt to recover Starship on land next year.

Orion’s other options. The Orion spacecraft and Space Launch System rocket have been attached at the hip for the better part of two decades. The big rocket lifts, the smaller spacecraft flies, and Congress keeps the money rolling in. But now there are signs that the twain may, in the not-too-distant future, split, Ars reports. This is because Lockheed Martin has begun to pivot toward a future in which the Orion spacecraft—thanks to increasing reusability, a focus on cost, and openness to flying on different rockets—fits into commercial space applications. In interviews, company officials said that if NASA wanted to buy Orion missions as a “service,” rather than owning and operating the spacecraft, they were ready to work with the space agency.

Staying power … This represents a significant change. Since the US Congress called for the creation of the Space Launch System rocket a decade and a half ago, Orion and this rocket have been discussed in tandem, forming the backbone of an expendable architecture that would launch humans to the Moon and return them to Earth inside Orion. But time is running out for the uber-expensive SLS rocket, with differing proposals from the Trump administration and Congress to terminate the program after either two or perhaps four more flights. This appears to be one reason Lockheed is exploring alternative launch vehicles for Orion. If the spacecraft is going to be competitive on price, it needs a rocket that does not cost more than $2 billion per launch. Any near-term plan to send astronauts to the Moon will still require Orion.

Doubling up at Vandenberg. The Department of the Air Force has approved SpaceX’s plans to launch up to 100 missions per year from Vandenberg Space Force Base in California, Ars reports. This would continue the tectonic turnaround at the spaceport on California’s Central Coast. Five years ago, Vandenberg hosted just a single orbital launch. This year’s number stands at 51 orbital flights, or 53 launches if you count a pair of Minuteman missile tests, the most in a single calendar year at Vandenberg since the early 1970s. Military officials have now authorized SpaceX to double its annual launch rate at Vandenberg from 50 to 100, with up to 95 missions using the Falcon 9 rocket and up to five launches of the larger Falcon Heavy.

No big rush … There’s more to the changes at Vandenberg than launching additional rockets. The authorization gives SpaceX the green light to redevelop Space Launch Complex 6 (SLC-6) to support Falcon 9 and Falcon Heavy missions. SpaceX plans to demolish unneeded structures at SLC-6 (pronounced “Slick 6”) and construct two new landing pads for Falcon boosters on a bluff overlooking the Pacific just south of the pad. SLC-6 would become the West Coast home for Falcon Heavy, but SpaceX currently has no confirmed contracts to fly the heavy-lifter from Vandenberg.

Next three launches

Oct. 18: Falcon 9 | Starlink 11-19 | Vandenberg Space Force Base, California | 23: 46 UTC

Oct. 19: Kinetica 1 | Unknown Payload | Jiuquan Satellite Launch Center, China | 03: 30 UTC

Oct. 19: Falcon 9 | Starlink 10-17 | Cape Canaveral Space Force Station, Florida | 14: 52 UTC

Stephen Clark is a space reporter at Ars Technica, covering private space companies and the world’s space agencies. Stephen writes about the nexus of technology, science, policy, and business on and off the planet.

Rocket Report: China launches with no advance warning; Europe’s drone ship Read More »

RFK Jr.’s MAHA wants to make chemtrail conspiracy theories great again

health / DJ Henderson / October 17, 2025

A prominent voice in the Make America Healthy Again movement is pushing for health secretary and anti-vaccine activist Robert F. Kennedy Jr. to make the topic of chemtrail conspiracy theories a federal priority, according to a report by KFF News.

KFF obtained a memo, written by MAHA influencer Gray Delany in July, presenting the topic to Calley Means, a White House health advisor. The memo lays out a series of unsubstantiated and far-fetched claims that academic researchers and federal agencies are secretively spreading toxic substances from airplanes, poisoning Americans, and spurring large-scale weather events, such as the devastating flooding in Texas last summer.

“It is unconscionable that anyone should be allowed to spray known neurotoxins and environmental toxins over our nation’s citizens, their land, food and water supplies,” Delany writes in the memo.

Daniel Swain, a climate scientist at the University of California, Los Angeles, told KFF that the memo presents claims that are false and, in some cases, physically impossible. “That is a pretty shocking memo,” he said. “It doesn’t get more tinfoil hat. They really believe toxins are being sprayed.”

Delany ends the memo with recommendations for federal agencies: form a joint task force to address this alleged geoengineering, host a roundtable on the topic, include the topic in the MAHA commission report, and publicly address the health and environmental harms.

It remains unclear if Kennedy, Means, or federal agencies are following up on Delany’s suggestions. Department of Health and Human Services spokesperson Emily Hilliard told KFF that “HHS does not comment on future or potential policy decisions and task forces.”

However, one opportunity has already been missed: The MAHA Commission released its “Make Our Children Healthy Again” report on September 9, along with a strategy document. Neither document mentions any of the topics raised in Delany’s memo.

RFK Jr.’s MAHA wants to make chemtrail conspiracy theories great again Read More »

Believing misinformation is a “win” for some people, even when proven false

authoritarianism, misinformation, psychology, Science, syndication / DJ Henderson / October 15, 2025

Why people endorse misinformation

Our findings highlight the limits of countering misinformation directly, because for some people, literal truth is not the point.

For example, President Donald Trump incorrectly claimed in August 2025 that crime in Washington, DC, was at an all-time high, generating countless fact-checks of his premise and think pieces about his dissociation from reality.

But we believe that to someone with a symbolic mindset, debunkers merely demonstrate that they’re the ones reacting and are therefore weak. The correct information is easily available but is irrelevant to someone who prioritizes a symbolic show of strength. What matters is signaling one isn’t listening and won’t be swayed.

In fact, for symbolic thinkers, nearly any statement should be justifiable. The more outlandish or easily disproved something is, the more powerful one might seem when standing by it. Being an edgelord—a contrarian online provocateur—or outright lying can, in their own odd way, appear “authentic.”

Some people may also view their favorite dissembler’s claims as provocative trolling, but, given the link between this mindset and authoritarianism, they want those far-fetched claims acted on anyway. The deployment of National Guard troops to Washington, for example, can be the desired end goal, even if the offered justification is a transparent farce.

Is this really 5-D chess?

It is possible that symbolic, but not exactly true, beliefs have some downstream benefit, such as serving as negotiation tactics, loyalty tests, or a fake-it-till-you-make-it long game that somehow, eventually, becomes a reality. Political theorist Murray Edelman, known for his work on political symbolism, noted that politicians often prefer scoring symbolic points over delivering results—it’s easier. Leaders can offer symbolism when they have little tangible to provide.

Randy Stein is associate professor of marketing at California State Polytechnic University, Pomona and Abraham Rutchick is professor of psychology at California State University, Northridge.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Believing misinformation is a “win” for some people, even when proven false Read More »

Feds seize $15 billion from alleged forced labor scam built on “human suffering”

Bitcoin, Biz & IT, federal indictments, online scams, pig butchering, Policy / DJ Henderson / October 15, 2025

Federal prosecutors have seized $15 billion from the alleged kingpin of an operation that used imprisoned laborers to trick unsuspecting people into making investments in phony funds, often after spending months faking romantic relationships with the victims.

Such “pig butchering” scams have operated for years. They typically work when members of the operation initiate conversations with people on social media and then spend months messaging them. Often, the scammers pose as attractive individuals who feign romantic interest for the victim.

Forced labor, phone farms, and human suffering

Eventually, conversations turn to phony investment funds with the end goal of convincing the victim to transfer large amounts of bitcoin. In many cases, the scammers are trafficked and held against their will in compounds surrounded by fences and barbed wire.

On Tuesday, federal prosecutors unsealed an indictment against Chen Zhi, the founder and chairman of a multinational business conglomerate based in Cambodia. It alleged that Zhi led such a forced-labor scam operation, which, with the help of unnamed co-conspirators, netted billions of dollars from victims.

“The defendant CHEN ZHI and his co-conspirators designed the compounds to maximize profits and personally ensured that they had the necessary infrastructure to reach as many victims as possible,” prosecutors wrote in the court document, filed in US District Court for the Eastern District of New York. The indictment continued:

For example, in or about 2018, Co-Conspirator-1 was involved in procuring millions of mobile telephone numbers and account passwords from an illicit online marketplace. In or about 2019, Co-Conspirator-3 helped oversee construction of the Golden Fortune compound. CHEN himself maintained documents describing and depicting “phone farms,” automated call centers used to facilitate cryptocurrency investment fraud and other cybercrimes, including the below image:

Credit: Justice Department

Prosecutors said Zhi is the founder and chairman of Prince Group, a Cambodian corporate conglomerate that ostensibly operated dozens of legitimate business entities in more than 30 countries. In secret, however, Zhi and top executives built Prince Group into one of Asia’s largest transnational criminal organizations. Zhi’s whereabouts are unknown.

Feds seize $15 billion from alleged forced labor scam built on “human suffering” Read More »

Google’s Photoshop-killer AI model is coming to search, Photos, and NotebookLM

AI, Artificial Intelligence, generative ai, Google, google notebookLM, google photos, search, Tech / DJ Henderson / October 14, 2025

NotebookLM added a video overview feature several months back, which uses AI to generate a video summary of the content you’ve added to the notebook. The addition of Nano Banana to NotebookLM is much less open-ended. Instead of entering prompts to edit images, NotebookLM has a new set of video styles powered by Nano Banana, including whiteboard, anime, retro print, and more. The original style is still available as “Classic.”

My favorite video.

NotebookLM’s videos are still somewhat limited, but this update adds a second general format. You can now choose “Brief” in addition to “Explainer,” with the option to add prompts that steer the video in the right direction. Although, that’s not a guarantee, as this is still generative AI. At least the style should be more consistent with the addition of Nano Banana.

The updated image editor is also coming to Google Photos, but Google doesn’t have a firm timeline. Google claims that its Nano Banana model is a “major upgrade” over its previous image-editing model. Conversational editing was added to Photos last month, but it’s not the Nano Banana model that has impressed testers over the summer. Google says that Nano Banana will arrive in the Photos app in the next few weeks, which should make those conversational edits much less frustrating.

Google’s Photoshop-killer AI model is coming to search, Photos, and NotebookLM Read More »

Trump admin fires more health employees amid government shutdown

health, HHS, lay off, Russ Vought / DJ Henderson / October 11, 2025

Questionable cull

Today’s layoffs are the work of White House Budget Director Russell Vought, a lead creator of the Project 2025 playbook, which planned a massive reduction in the federal workforce. In a post on X earlier today, Vought announced that the terminations “have begun.”

But as The Washington Post has previously reported, senior government officials have warned that Vought’s layoffs amid a shutdown are likely illegal, running afoul of the Antideficiency Act. The law forbids the government from incurring new expenses during a shutdown, and the process of laying employees off—which includes severance packages—does just that.

Federal employment lawyers told the Post that the move is almost certainly illegal for a second reason: Under federal regulations, a shutdown-driven lapse in funding does not count as one of the reasons federal employees can be terminated.

Last week, the American Federation of Government Employees and other unions representing federal workers filed a lawsuit over threats that the Trump administration would try to lay off workers during the shutdown.

In a statement today, AFGE National President Everett Kelley said, “It is disgraceful that the Trump administration has used the government shutdown as an excuse to illegally fire thousands of workers who provide critical services to communities across the country.”

“AFGE is currently challenging President Trump’s illegal, unprecedented abuse of power, and we will not stop fighting until every reduction-in-force notice is rescinded,” Kelley said.

Trump admin fires more health employees amid government shutdown Read More »

Putin OKs plan to turn Russian spacecraft into flying billboards

advertising, launch, roscosmos, russia, Science, Space / DJ Henderson / October 11, 2025

These are tough times for Russia’s civilian space program. In the last few years, Russia has cut back on the number of Soyuz crew missions it is sending to the International Space Station, and a replacement for the nearly 60-year-old Soyuz spacecraft remains elusive.

While the United States and China are launching more space missions than ever before, Russia’s once-dominant launch cadence is on a downhill slide.

Russia’s access to global markets dried up after Russian President Vladimir Putin launched the country’s invasion of Ukraine in February 2022. The fallout from the invasion killed several key space partnership between Russia and Europe. Russia’s capacity to do new things in space seems to be focused on military programs like anti-satellite weapons.

The Roscosmos State Corporation for Space Activities, Russia’s official space agency, may have a plan to offset the decline. Late last month, Putin approved changes to federal laws governing advertising and space activities to “allow for the placement of advertising on spacecraft,” Roscosmos posted on its official Telegram account.

We’ve seen this before

The Russian State Duma, dominated by Putin loyalists, previously approved the amendments.

“According to the amendments, Roscosmos has been granted the right, effective January 1, 2026, to place advertising on space objects owned by both the State Corporation itself and federally,” Roscosmos said. “The amendments will create a mechanism for attracting private investment in Russian space exploration and reduce the burden on the state budget.”

The law requires that advertising symbols not affect spacecraft safety. The Russian government said it will establish a fee structure for advertising on federally owned space objects.

Roscosmos didn’t say this, but advertisers eligible for the offer will presumably be limited to Russia and its allies. Any ads from the West would likely violate sanctions.

Rocket-makers have routinely applied decals, stickers, and special paint jobs to their vehicles. This is a particularly popular practice in Russia. Usually, these logos represent customers and suppliers. Sometimes they honor special occasions, like the 60th anniversary of the first human spaceflight mission by Soviet cosmonaut Yuri Gagarin and the 80th anniversary of the end of World War II.

Putin OKs plan to turn Russian spacecraft into flying billboards Read More »

Bose SoundTouch home theater systems regress into dumb speakers Feb. 18

audio, bose, Smart Home, Sonos, Speakers, Tech / DJ Henderson / October 11, 2025

Bose will brick key features of its SoundTouch Wi-Fi speakers and soundbars soon. On Thursday, Bose informed customers that as of February 18, 2026, it will stop supporting the devices, and the devices’ cloud-based features, including the companion app, will stop working.

The SoundTouch app enabled numerous capabilities, including integrating music services, like Spotify and TuneIn, and the ability to program multiple speakers in different rooms to play the same audio simultaneously.

Bose has also said that some saved presets won’t work and that users won’t be able to change saved presets once the app is gone.

Additionally, Bose will stop providing security updates for SoundTouch devices.

The Framingham, Massachusetts-headquartered company noted to customers that the speakers will continue being able to play audio from a device connected via AUX or HDMI. Wireless playback will still work over Bluetooth; however, Bluetooth is known to introduce more latency than Wi-Fi connections.

Affected customers can trade in their SoundTouch product for a credit worth up to $200.

In its notice sent to customers this week, Bose provided minimal explanation for end-of-life-ing its pricey SoundTouch speakers, saying:

Bose SoundTouch systems were introduced into the market in 2013. Technology has evolved since then, and we’re no longer able to sustain the development and support of the cloud infrastructure that powers this older generation of products. We remain committed to creating new listening experiences for our customers built on modern technologies.

Ars Technica has reached out to Bose for comment.

“Really disgusted”

Bose launched SoundTouch with three speakers ranging from $399 to $699. The company marketed the wireless home audio system as a way to extend high-quality sound throughout the home using Wi-Fi-connected speakers.

In 2015, Bose expanded the lineup with speakers ranging from $200 to $400 and soundbars and home theater systems ranging from $1,100 to $1,500.

By 2020, however, Bose was distancing itself from SoundTouch. It informed customers that it was “discontinuing sales of some SoundTouch products” but said it was “committed” to supporting the “SoundTouch app and product software for the foreseeable future.” Apparently, Bose couldn’t see beyond the next five years.

Bose SoundTouch home theater systems regress into dumb speakers Feb. 18 Read More »

OpenAI will stop saving most ChatGPT users’ deleted chats

AI, Policy / DJ Henderson / October 10, 2025

Moving forward, all of the deleted and temporary chats that were previously saved under the preservation order will continue to be accessible to news plaintiffs, who are looking for examples of outputs infringing their articles or attributing misinformation to their publications.

Additionally, OpenAI will continue monitoring certain ChatGPT accounts, saving deleted and temporary chats of any users whose domains have been flagged by news organizations since they began searching through the data. If news plaintiffs flag additional domains during future meetings with OpenAI, more accounts could be roped in.

Ars could not immediately reach OpenAI or the Times’ legal team for comment.

The dispute with news plaintiffs continues to heat up beyond the battle over user logs, most recently with co-defendant Microsoft pushing to keep its AI companion Copilot out of the litigation.

The stakes remain high for both sides. News organizations have alleged that ChatGPT and other allegedly copyright-infringing tools threaten to replace them in their market while potentially damaging their reputations by attributing false information to them.

OpenAI may be increasingly pressured to settle the lawsuit, and not by news organizations but by insurance companies that won’t provide comprehensive coverage for their AI products with multiple potentially multibillion-dollar lawsuits pending.

OpenAI will stop saving most ChatGPT users’ deleted chats Read More »

AI models can acquire backdoors from surprisingly few malicious documents

AI, AI research, AI security, AI vulnerabilities, alan turing institute, Anthropic, backdoor attacks, Biz & IT, data poisoning, fine tuning, LLM security, machine learning, model safety, pretraining, training data, UK AI Security Institute / DJ Henderson / October 10, 2025

Fine-tuning experiments with 100,000 clean samples versus 1,000 clean samples showed similar attack success rates when the number of malicious examples stayed constant. For GPT-3.5-turbo, between 50 and 90 malicious samples achieved over 80 percent attack success across dataset sizes spanning two orders of magnitude.

Limitations

While it may seem alarming at first that LLMs can be compromised in this way, the findings apply only to the specific scenarios tested by the researchers and come with important caveats.

“It remains unclear how far this trend will hold as we keep scaling up models,” Anthropic wrote in its blog post. “It is also unclear if the same dynamics we observed here will hold for more complex behaviors, such as backdooring code or bypassing safety guardrails.”

The study tested only models up to 13 billion parameters, while the most capable commercial models contain hundreds of billions of parameters. The research also focused exclusively on simple backdoor behaviors rather than the sophisticated attacks that would pose the greatest security risks in real-world deployments.

Also, the backdoors can be largely fixed by the safety training companies already do. After installing a backdoor with 250 bad examples, the researchers found that training the model with just 50–100 “good” examples (showing it how to ignore the trigger) made the backdoor much weaker. With 2,000 good examples, the backdoor basically disappeared. Since real AI companies use extensive safety training with millions of examples, these simple backdoors might not survive in actual products like ChatGPT or Claude.

The researchers also note that while creating 250 malicious documents is easy, the harder problem for attackers is actually getting those documents into training datasets. Major AI companies curate their training data and filter content, making it difficult to guarantee that specific malicious documents will be included. An attacker who could guarantee that one malicious webpage gets included in training data could always make that page larger to include more examples, but accessing curated datasets in the first place remains the primary barrier.

Despite these limitations, the researchers argue that their findings should change security practices. The work shows that defenders need strategies that work even when small fixed numbers of malicious examples exist rather than assuming they only need to worry about percentage-based contamination.

“Our results suggest that injecting backdoors through data poisoning may be easier for large models than previously believed as the number of poisons required does not scale up with model size,” the researchers wrote, “highlighting the need for more research on defences to mitigate this risk in future models.”

AI models can acquire backdoors from surprisingly few malicious documents Read More »

How Easter Island’s giant statues “walked” to their final platforms

anthropology, Archaeology, easter island, experimental archaeology, Physics, rapa nui, Science / DJ Henderson / October 9, 2025

Workers with ropes could make the moai “walk” in zig-zag motion along roads tailor-made for the purpose.

Easter Island is famous for its giant monumental statues, called moai, built some 800 years ago and typically mounted on platforms called ahu. Scholars have puzzled over the moai on Easter Island for decades, pondering their cultural significance, as well as how a Stone Age culture managed to carve and transport statues weighing as much as 92 tons. One hypothesis, championed by archaeologist Carl Lipo of Binghamton University, among others, is that the statues were transported in a vertical position, with workers using ropes to essentially “walk” the moai onto their platforms.

The oral traditions of the people of Rapa Nui certainly include references to the moai “walking” from the quarry to their platforms, such as a song that tells of an early ancestor who made the statues walk. While there have been rudimentary field tests showing it might have been possible, the hypothesis has also generated a fair amount of criticism. So Lipo has co-authored a new paper published in the Journal of Archaeological Science offering fresh experimental evidence of “walking” moai, based on 3D modeling of the physics and new field tests to recreate that motion.

The first Europeans arrived in the 17th century and found only a few thousand inhabitants on the tiny island (just 14 by 7 miles across) thousands of miles away from any other land. In order to explain the presence of so many moai, the assumption has been that the island was once home to tens of thousands of people. But Lipo thought perhaps the feat could be accomplished with fewer workers. In 2012, Lipo and his colleague, Terry Hunt of the University of Arizona, showed that you could transport a 10-foot, 5-ton moai a few hundred yards with just 18 people and three strong ropes by employing a rocking motion.

In 2018, Lipo followed up with an intriguing hypothesis for how the islanders placed red hats on top of some moai; those can weigh up to 13 tons. He suggested the inhabitants used ropes to roll the hats up a ramp. Lipo and his team later concluded (based on quantitative spatial modeling) that the islanders likely chose the statues’ locations based on the availability of fresh water sources, per a 2019 paper in PLOS One.

The 2012 experiment demonstrated proof of principle, so why is Lipo revisiting it now? “I always felt that the [original] experiment was disconnected to some degree of theory—that we didn’t have particular expectations about numbers of people, rate of transport, road slope that could be walked, and so on,” Lipo told Ars. There were also time constraints because the attempt was being filmed for a NOVA documentary.

“That experiment was basically a test to see if we could make it happen or not,” he explained. “Fortunately, we did, and our joy in doing so is pretty well represented by our hoots and hollers when it started to walk with such limited efforts. Some of the limitation of the work was driven by the nature of TV. [The film crew] just wanted us—in just a day and half—to give it a shot. It was 4: 30 on the last day when it finally worked so we really didn’t get a lot of time to explore variability. We also didn’t have any particular predictions to test.”

Example of a road moai that fell and was abandoned after an attempt to re-erect it by excavating under its base, leaving it partially buried at an angle. Credit: Carl Lipo

This time around, “We wanted to explore a bit of the physics: to show that what we did was pretty easily predicted by the physical properties of the moai—its shape, size, height, number of people on ropes, etc.—and that our success in terms of team size and rate of walking was consistent with predictions,” said Lipo. “This enables us to address one of the central critiques that always comes up: ‘Well, you did this with a 5-ton version that was 10 feet tall, but it would never work with a 30-ft-tall version that weighs 30 tons or more.'”

All about that base

You can have ahu (platforms) without moai (statues) and moai without ahu, usually along the roads leading to ahu; they were likely being transported and never got to their destination. Lipo and Hunt have amassed a database of 962 moai across the island, compiled through field surveys and photogrammetric documentation. They were particularly interested in 62 statues located along ancient transport roads that seemed to have been abandoned where they fell.

Their analysis revealed that these road moai had significantly wider bases relative to shoulder width, compared to statues mounted on platforms. This creates a stable foundation that lowers the center of mass so that the statue is more conducive to the side-to-side motion of walking transport without toppling over. Platform statues, by contrast, have shoulders wider than the base for a more top-heavy configuration.

The road moai also have a consistent and pronounced forward lean of between 6 degrees to 15 degrees from the vertical position, which moves the center of mass close to or just beyond the base’s front edge. Lipo and Hunt think this was due to careful engineering, not coincidence. It’s not conducive to stable vertical display but it is a boon during walking transport, because the forward lean causes the statue to fall forward when tilted laterally, with the rounded front base edge serving as a crucial pivot point. So every lateral rocking motion results in a forward “step.”

Per the authors, there is strong archaeological evidence that carvers modified the statues once they arrived at their platform destinations, modifying the base to eliminate the lean by removing material from the front. This shifted the center of mass over the base area for a stable upright position. The road moai even lack the carved eye sockets designed to hold white coral eyes with obsidian or red scoria for pupils—a final post-transport step once the statues had been mounted on their platforms.

Based on 3D modeling, Lipo and his team created a precisely scaled replica of one of the road moai, weighing 4.35 metric tons with the same proportions and mass distribution of the original statue. “Of course, we’d love to build a 30-foot-tall version, but the physical impossibility of doing so makes it a challenging task, nor is it entirely necessary,” said Lipo. “Through physics, we can now predict how many people it would take and how it would be done. That is key.”

Lipo's team created 3D models of moai to determine the unique characteristics that made them able to be — Lipo’s team created 3D models of moai to determine the unique characteristics that made them able to be “walked” across Rapa Nui. Credit: Carl Lipo

The new field trials required 18 people, four on each lateral rope and 10 on a rear rope, to achieve the side-to-side walking motion, and they were efficient enough in coordinating their efforts to move the statue forward 100 meters in just 40 minutes. That’s because the method operates on basic pendulum dynamics, per the authors, which minimizes friction between the base and the ground. It’s also a technique that exploits the gradual build-up of amplitude, which “suggests a sophisticated understanding of resonance principles,” Lipo and Hunt wrote.

So the actual statues could have been moved several kilometers over the course of weeks with only modest-sized crews of between 20-50 people, i.e., roughly the size of an extended family or “small lineage group” on Easter Island. Once the crew gets the statue rocking side to side—which can require between 15 to 60 people, depending on the size and weight of the moai—the resulting oscillation only needs minimal energy input from a smaller team of rope handlers to maintain that motion. They mostly provide guidance.

Lipo was not the first to test the walking hypothesis. Earlier work includes that of Czech experimental archaeologist Pavel Pavel, who conducted similar practical experiments on Easter Island in the 1980s after being inspired by Thor Heyerdahl’s Kon Tiki. (Heyerdahl even participated in the experiments.) Pavel’s team was able to demonstrate a kind of “shuffling” motion, and he concluded that just 16 men and one leader were sufficient to transport the statues.

Per Lipo and Hunt, Pavel’s demonstration didn’t result in broad acceptance of the walking hypothesis because it still required a huge amount of effort to tilt the statue, producing more of a twisting motion rather than efficient forward movement. This would only have moved a large statue 100 meters a day under ideal conditions. The base was also likely to be damaged from friction with the ground. Lipo and Hunt maintain this is because Pavel (and others who later tried to reproduce his efforts) used the wrong form of moai for those earlier field tests: those erected on the platforms, already modified for vertical stability and permanent display, and not the road moai with shapes more conducive to vertical transport.

“Pavel deserves recognition for taking oral traditions seriously and challenging the dominant assumption of horizontal transport, a move that invited ridicule from established scholars,” Lipo and Hunt wrote. “His experiments suggested that vertical transport was feasible and consistent with cultural memory. Our contribution builds on this by showing that ancestral engineers intentionally designed statues for walking. Those statues were later modified to stand erect on ceremonial platforms, a transformation that effectively erased the morphological features essential for movement.”

The evidence of the roadways

Lipo and Hunt also analyzed the roadways, noting that these ancient roadbeds had concave cross sections that would have been problematic for moving the statues horizontally using wooden rollers or frames perpendicular to those roads. But that concave shape would help constrain rocking movement during vertical transport. And the moai roads were remarkably level with slopes of, on average, 2–3 percent. For the occasional steeper slopes, such as walking a moai up a ramp to the top of an ahu, Lipo and Hunt’s field experiments showed that these could be navigated successfully through controlled stepping.

Furthermore, the distribution pattern of the roadways is consistent with the road moai being left due to mechanical failure. “Arguments that the moai were placed ceremonially in preparation for quarrying have become more common,” said Lipo. “The algorithm there is to claim that positions are ritual, without presenting anything that is falsifiable. There is no reason why the places the statues fell due to mechanical reasons couldn’t later become ‘ritual,’ in the same way that everything on the island could be claimed to be ritual—a circular argument. But to argue that they were placed there purposefully for ritual purposes demands framing the explanation in a way that is falsifiable.”

Schematic representation of the moai transport method using coordinated rope pulling to achieve a “walking” motion. Credit: Carl Lipo and Terry Hunt, 2025

“The only line of evidence that is presented in this way is the presence of ‘platforms’ that were found beneath the base of one moai, which is indeed intriguing,” Lipo continued. “However, those platforms can be explained in other ways, given that the moai certainly weren’t moved from the quarry to the ahu in one single event. They were paused along the way, as is clear from the fact that the roads appear to have been constructed in segments with different features. Their construction appears to be part of the overall transport process.”

Lipo’s work has received a fair share of criticism from other scholars over the years, and his and Hunt’s paper includes a substantial section rebutting the most common of those critiques. “Archaeologists tend to reject (in practice) the idea that the discipline can construct cumulative knowledge,” said Lipo. “In the case of moai transport, we’ve strived to assemble as much empirical evidence as possible and have forwarded an explanation that best accounts for what we can observe. Challenges to these ideas, however, do not come from additional studies with new data but rather just new assertions.”

“This leads the public to believe that we (as a discipline) can never really figure anything out and are always going to be a speculative enterprise, spinning yarns and arguing with each other,” Lipo continued. “With the erosion of trust in science, this is fairly catastrophic to archaeology as a whole but also the whole scientific enterprise. Summarizing the results in the way we do here is an attempt to point out that we can build falsifiable accounts and can make contributions to cumulative knowledge that have empirical consequences—even with something as remarkable as the transport of moai.”

Experimental archaeology is a relatively new field that some believe could be the future of archaeology. “I think experimental archaeology has potential when it’s tied to physics and chemistry,” said Lipo. “It’s not just recreating something and then arguing it was done in the same way in the past. Physics and chemistry are our time machines, allowing us to explain why things are the way they are in the present in terms of the events that occurred in the past. The more we can link the theory needed to explain the present, the better we can explain the past.”

DOI: Journal of Archaeological Science, 2025. 10.1016/j.jas.2025.106383 (About DOIs).

Jennifer is a senior writer at Ars Technica with a particular focus on where science meets culture, covering everything from physics and related interdisciplinary topics to her favorite films and TV series. Jennifer lives in Baltimore with her spouse, physicist Sean M. Carroll, and their two cats, Ariel and Caliban.

How Easter Island’s giant statues “walked” to their final platforms Read More »

Bank of England warns AI stock bubble rivals 2000 dotcom peak

AI, AI stocks, alphabet, Amazon, andrew bailey, Apple, bank of england, Biz & IT, Donald Trump, dotcom bubble, Federal Reserve, financial stability, Google, Lisa Cook, machine learning, mark zuckerberg, market correction, Meta, microsoft, NVIDIA, UK economy / DJ Henderson / October 9, 2025

Share valuations based on past earnings have also reached their highest levels since the dotcom bubble 25 years ago, though the BoE noted they appear less extreme when based on investors’ expectations for future profits. “This, when combined with increasing concentration within market indices, leaves equity markets particularly exposed should expectations around the impact of AI become less optimistic,” the central bank said.

Toil and trouble?

The dotcom bubble offers a potentially instructive parallel to our current era. In the late 1990s, investors poured money into Internet companies based on the promise of a transformed economy, seemingly ignoring whether individual businesses had viable paths to profitability. Between 1995 and March 2000, the Nasdaq index rose 600 percent. When sentiment shifted, the correction was severe: the Nasdaq fell 78 percent from its peak, reaching a low point in October 2002.

Whether we’ll see the same thing or worse if an AI bubble pops is mere speculation at this point. But similar to the early 2000s, the question about today’s market isn’t necessarily about the utility of AI tools themselves (the Internet was useful, afterall, despite the bubble), but whether the amount of money being poured into the companies that sell them is out of proportion with the potential profits those improvements might bring.

We don’t have a crystal ball to determine when such a bubble might pop, or even if it is guaranteed to do so, but we’ll likely continue to see more warning signs ahead if AI-related deals continue to grow larger and larger over time.

Bank of England warns AI stock bubble rivals 2000 dotcom peak Read More »

Author name: DJ Henderson