Semua Kabar

Canada is the only G7 nation without a launch program. Quebec wants to do something about that.

Welcome to Edition 7.48 of the Rocket Report! The shock of last week's public spat between President Donald Trump and SpaceX founder Elon Musk has worn off, and Musk expressed regret for some of his comments going after Trump on social media. Musk also backtracked from his threat to begin decommissioning the Dragon spacecraft, currently the only way for the US government to send people to the International Space Station. Nevertheless, there are many people who think Musk's attachment to Trump could end up putting the US space program at risk, and I'm not convinced that danger has passed.

As always, wewelcome 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.

Quebec invests in small launch company.The government of Quebec will invest CA$10 million ($7.3 million) into a Montreal-area company that is developing a system to launch small satellites into space,The Canadian Press reports. Quebec Premier François Legault announced the investment into Reaction Dynamics at the company's facility in Longueuil, a Montreal suburb. The province's economy minister, Christine Fréchette, said the investment will allow the company to begin launching microsatellites into orbit from Canada as early as 2027.

Joining its peers… Canada is the only G7 nation without a domestic satellite launch capability, whether it's through an independent national or commercial program or through membership in the European Space Agency, which funds its own rockets. The Canadian Space Agency has long eschewed any significant spending on developing a Canadian satellite launcher, and a handful of commercial launch startups in Canada haven't gotten very far. Reaction Dynamics was founded in 2017 by Bachar Elzein, formerly a researcher in multiphase and reactive flows at École Polytechnique de Montréal, where he specialized in propulsion and combustion dynamics. Reaction Dynamic plans to launch its first suborbital rocket later this year, before attempting an orbital flight with its Aurora rocket as soon as 2027. (submitted by Joey S-IVB)

Another year, another delay for Themis.The European Space Agency’s Themis program has suffered another setback, with the inaugural flight of its reusable booster demonstrator now all but certain to slip to 2026,European Spaceflight reports. It has been nearly six years since the European Space Agency kicked off the Themis program to develop and mature key technologies for future reusable rocket stages. Themis is analogous to SpaceX's Grasshopper reusable rocket prototype tested more than a decade ago, with progressively higher hop tests to demonstrate vertical takeoff and vertical landing techniques. When the program started, an initial hop test of the first Themis demonstrator was expected to take place in 2022.

Tethered to terra firma… ArianeGroup, which manufactures Europe's Ariane rockets, is leading the Themis program under contract to ESA, which recently committed an additional 230 million euros ($266 million) to the effort. This money is slated to go toward the development of a single-engine variant of the Themis program, continued development of the rocket's methane-fueled engine, and upgrades to a test stand at ArianeGroup's propulsion facility in Vernon, France. Two months ago, an official update on the Themis program suggested the first Themis launch campaign would begin before the end of the year. Citing sources close to the program, European Spaceflight reports the first Themis integration tests at the Esrange Space Center in Sweden are now almost certain to slip from late 2025 to 2026.

French startup tests a novel rocket engine.While Europe's large government-backed rocket initiatives face delays, the continent's space industry startups are moving forward on their own. One of these companies, a French startup named Alpha Impulsion, recently completed a short test-firing of an autophage rocket engine,European Spaceflight reports. These aren't your normal rocket engines that burn conventional kerosene, methane, or hydrogen fuel. An autophage engine literally consumes itself as it burns, using heat from the combustion process to melt its plastic fuselage and feed the molten plastic into the combustion chamber in a controlled manner. Alpha Impulsion called the May 27 ground firing a successful test of the "largest autophage rocket engine in the world."

So, why hasn't this been done before?… The concept of a self-consuming rocket engine sounds like an idea that's so crazy it just might work. But the idea remained conceptual from when it was first patented in 1938 until an autophage engine was fired in a controlled manner for the first time in 2018. The autophage design offers several advantages, including its relative simplicity compared to the complex plumbing of liquid and hybrid rockets. But there are serious challenges associated with autophage engines, including how to feed molten fuel into the combustion chamber and how to scale it up to be large enough to fly on a viable rocket.(submitted by trimeta and EllPeaTea)

Rocket trouble delays launch of private crew mission.A propellant leak in a Falcon 9 booster delayed the launch of a fourth Axiom Space private astronaut mission to the International Space Station this week,Space News reports. SpaceX announced the delay Tuesday, saying it needed more time to fix a liquid oxygen leak found in the Falcon 9 booster during inspections following a static-fire test Sunday. "Once complete–and pending Range availability–we will share a new launch date," the company stated. The Ax-4 mission will ferry four commercial astronauts, led by retired NASA commander Peggy Whitson, aboard a Dragon spacecraft to the ISS for an approximately 14-day stay. Whitson will be joined by crewmates from India, Poland, and Hungary.

Another problem, too… While SpaceX engineers worked on resolving the propellant leak on the ground, a leak of another kind in orbit forced officials to order a longer delay to the Ax-4 mission. In a statement Thursday, NASA said it is working with the Russian space agency to understand a "new pressure signature" in the space station's Russian service module. For several years, ground teams have monitored a slow air leak in the aft part of the service module, and NASA officials have identified it as a safety risk. NASA's statement on the matter was vague, only saying that cosmonauts on the station recently inspected the module's interior surfaces and sealed additional "areas of interest." The segment is now holding pressure, according to NASA. (submitted by EllPeaTea)

SpaceX tries something new with Falcon 9.With nearly 500 launches under its belt, SpaceX's Falcon 9 rocket isn't often up to new tricks. But the company tried something new following a launch on June 7 with a radio broadcasting satellite for SiriusXM. The Falcon 9's upper stage placed the SXM-10 satellite into an elongated, high-altitude transfer orbit, as is typical for payloads destined to operate in geosynchronous orbit more than 22,000 miles (nearly 36,000 kilometers) over the equator. When a rocket releases a satellite in this type of high-energy orbit, the upper stage has usually burned almost all of its propellant, leaving little fuel to steer itself back into Earth's atmosphere for a destructive reentry. This means these upper stages often remain in space for decades, becoming a piece of space junk that transits across the orbits of many other satellites.

Now, a solution… SpaceX usually deorbits rockets after they deploy payloads like Starlink satellites into low-Earth orbit, but deorbiting a rocket from a much higher geosynchronous transfer orbit is a different matter. "Last week, SpaceX successfully completed a controlled deorbit of the SiriusXM-10 upper stage after GTO payload deployment," wrote Jon Edwards, SpaceX's vice president of Falcon and Dragon programs. "While we routinely do controlled deorbits for LEO stages (e.g., Starlink), deorbiting from GTO is extremely difficult due to the high energy needed to alter the orbit, making this a rare and remarkable first for us. This was only made possible due to the hard work and brilliance of the Falcon GNC (guidance, navigation, and control) team and exemplifies SpaceX's commitment to leading in both space exploration and public safety."

New Glenn gets a tentative launch date.Five months have passed since Blue Origin's New Glenn rocket made its mostly successful debut in January. At one point, the company targeted "late spring" for the second launch of the rocket. However, on Monday, Blue Origin's CEO, Dave Limp, acknowledged on social media that the rocket's next flight will now no longer take place until at least August 15,Ars reports. Although he did not say so, this may well be the only other New Glenn launch this year. The mission, with an undesignated payload, will be named "Never Tell Me the Odds," due to the attempt to land the booster. "One of our key mission objectives will be to land and recover the booster,"Limp wrote. "This will take a little bit of luck and a lot of excellent execution. We’re on track to produce eight GS2s [second stages] this year, and the one we’ll fly on this second mission was hot-fired in April."

Falling short…Before 2025 began, Limp set expectations alongside Blue Origin founder Jeff Bezos: New Glenn would launch eight times this year. That's not going to happen. It's common for launch companies to take a while ramping up the flight rate for a new rocket, but Bezos told Ars in January that his priority for Blue Origin this year was to hit a higher cadence with New Glenn. Elon Musk's rift with President Donald Trump could open a pathway for Blue Origin to capture more government business if the New Glenn rocket is able to establish a reliable track record. Meanwhile, Limp told Blue Origin employees last month that Jarrett Jones, the manager running the New Glenn program, is taking a sabbatical. Although it appears Jones' leave may have been planned, the timing is curious.

Making way for Starship at Cape Canaveral.The US Air Force is moving closer to authorizing SpaceX to move into one of the largest launch pads at Cape Canaveral Space Force Station in Florida, with plans to use the facility for up to 76 launches of the company's Starship rocket each year,Ars reports. Adraft Environmental Impact Statement (EIS)released by the Department of the Air Force, which includes the Space Force, found SpaceX's planned use of Space Launch Complex 37 (SLC-37) at Cape Canaveral would have no significant negative impacts on local environmental, historical, social, and cultural interests. The Air Force also found SpaceX's plans at SLC-37 will have no significant impact on the company's competitors in the launch industry.

Bringing the rumble …SLC-37 was the previous home to United Launch Alliance's Delta IV rocket, which last flew from the site in April 2024, a couple of months after the militaryannounced SpaceX was interestedin using the launch pad. While it doesn't have a lease for full use of the launch site, SpaceX has secured a "right of limited entry" from the Space Force to begin preparatory work. This included theexplosive demolitionof the launch pad's Delta IV-era service towers and lightning masts Thursday, clearing the way for eventual construction of two Starship launch towers inside the perimeter of SLC-37. The new Starship launch towers at SLC-37 will join other properties in SpaceX's Starship empire, including nearby Launch Complex 39A at NASA's Kennedy Space Center, and SpaceX's privately owned facility at Starbase, Texas.

Preps continue for Starship Flight 10.Meanwhile, at Starbase, SpaceX is moving forward with preparations for the next Starship test flight, which could happen as soon as next month following three consecutive flights that fell short of expectations. This next launch will be the 10th full-scale test flight of Starship. Last Friday, June 6, SpaceX test-fired the massive Super Heavy booster designated to launch on Flight 10. All 33 of its Raptor engines ignited on the launch pad in South Texas. This is a new Super Heavy booster. On Flight 9 last month, SpaceX flew a reused Super Heavy booster that launched and was recovered on a flight in January.

FAA signs off on SpaceX investigation …The Federal Aviation Administration said Thursday it has closed the investigation into Starship Flight 8 in March, which spun out of control minutes after liftoff, showering debris along a corridor of ocean near the Bahamas and the Turks and Caicos Islands. "The FAA oversaw and accepted the findings of the SpaceX-led investigation," an agency spokesperson said. "The final mishap report cites the probable root cause for the loss of the Starship vehicle as a hardware failure in one of the Raptor engines that resulted in inadvertent propellant mixing and ignition. SpaceX identified eight corrective actions to prevent a reoccurrence of the event." SpaceX implemented the corrective actions prior to Flight 9 last month, when Starship progressed further into its mission before starting to tumble in space. It eventually reentered the atmosphere over the Indian Ocean. The FAA has mandated a fresh investigation into Flight 9, and that inquiry remains open.

June 13:Falcon 9 | Starlink 12-26 | Cape Canaveral Space Force Station, Florida | 15:21 UTC

June 14:Long March 2D | Unknown Payload | Jiuquan Satellite Launch Center, China | 07:55 UTC

June 16:Atlas V | Project Kuiper KA-02| Cape Canaveral Space Force Station, Florida | 17:25 UTC

Why requests for "no AI resurrections" will probably go ignored.

As artificial intelligence has advanced, AI tools have emerged to make it possible to easily create digital replicas of lost loved ones, which can be generated without the knowledge or consent of the person who died.

Trained on the data of the dead, these tools, sometimes called grief bots or AI ghosts, may be text-, audio-, or even video-based. Chatting provides what some mourners feel is a close approximation to ongoing interactions with the people they love most. But the tech remains controversial, perhaps complicating the grieving process while threatening to infringe upon the privacy of the deceased, whose data could still be vulnerable to manipulation or identity theft.

Because of suspected harms and perhaps a general repulsion to the idea of it, not everybody wants to become an AI ghost.

After a realistic video simulation was recently used to provide a murder victim's impact statement in court, Futurism summed up social media backlash,notingthat the use of AI was "just as unsettling as you think." And it's not the first time people have expressed discomfort with the growing trend. Last May, The Wall Street Journal conducted a readersurveyseeking opinions on the ethics of so-called AI resurrections. Responding, a California woman, Dorothy McGarrah, suggested there should be a way to prevent AI resurrections in your will.

"Having photos or videos of lost loved ones is a comfort. But the idea of an algorithm, which is as prone to generate nonsense as anything lucid, representing a deceased person’s thoughts or behaviors seems terrifying. It would be like generating digital dementia after your loved ones’ passing," McGarrah said. "I would very much hope people have the right to preclude their images being used in this fashion after death. Perhaps something else we need to consider in estate planning?"

For experts in estate planning, the question may start to arise as more AI ghosts pop up. But for now, writing "no AI resurrections" into a will remains a complicated process, experts suggest, and such requests may not be honored by all unless laws are changed to reinforce a culture of respecting the wishes of people who feel uncomfortable with the idea of haunting their favorite people through AI simulations.

Ars contacted several law associations to find out if estate planners are seriously talking about AI ghosts. Only the National Association of Estate Planners and Councils responded; it connected Ars to Katie Sheehan, an expert in the estate planning field who serves as a managing director and wealth strategist for Crestwood Advisors.

Sheehan told Ars that very few estate planners are prepared to answer questions about AI ghosts. She said not only does the question never come up in her daily work, but it's also "essentially uncharted territory for estate planners since AI is relatively new to the scene."

"I have not seen any documents drafted to date taking this into consideration, and I review estate plans for clients every day, so that should be telling," Sheehan told Ars.

Although Sheehan has yet to see a will attempting to prevent AI resurrection, she told Ars that there could be a path to make it harder for someone to create a digital replica without consent.

"You certainly could draft into a power of attorney (for use during lifetime) and a will (for use post death) preventing the fiduciary (attorney in fact or executor) from lending any of your texts, voice, image, writings, etc. to any AI tools and prevent their use for any purpose during life or after you pass away, and/or lay the ground rules for when they can and cannot be used after you pass away," Sheehan told Ars.

"This could also invoke issues with contract, property and intellectual property rights, and right of publicity as well if AI replicas (image, voice, text, etc.) are being used without authorization," Sheehan said.

And there are likely more protections for celebrities than for everyday people, Sheehan suggested.

"As far as I know, there is no law" preventing unauthorized non-commercial digital replicas, Sheehan said.

Widely adopted by states, the Revised Uniform Fiduciary Access to Digital Assets Act—which governs who gets access to online accounts of the deceased, like social media or email accounts—could be helpful but isn't a perfect remedy.

That law doesn’t directly "cover someone’s AI ghost bot, though it may cover some of the digital material some may seek to use to create a ghost bot," Sheehan said.

"Absent any law" blocking non-commercial digital replicas, Sheehan expects that people's requests for "no AI resurrections" will likely "be dealt with in the courts and governed by the terms of one’s estate plan, if it is addressed within the estate plan."

Those potential fights seemingly could get hairy, as "it may be some time before we get any kind of clarity or uniform law surrounding this," Sheehan suggested.

In the future, Sheehan said, requests prohibiting digital replicas may eventually become "boilerplate language in almost every will, trust, and power of attorney," just as instructions on digital assets are now.

As "all things AI become more and more a part of our lives," Sheehan said, "some aspects of AI and its components may also be woven throughout the estate plan regularly."

"But we definitely aren’t there yet," she said. "I have had zero clients ask about this."

Whether loved ones would—or even should—respect requests blocking digital replicas appears to be debatable. But at least one person who built a grief bot wished he'd done more to get his dad's permission before moving forward with his own creation.

A computer science professor at the University of Washington Bothell, Muhammad Aurangzeb Ahmad, was one of the earliest AI researchers to create a grief bot more than a decade ago after his father died. He built the bot to ensure that his future kids would be able to interact with his father after seeing how incredible his dad was as a grandfather.

When Ahmad started his project, there was no ChatGPT or other advanced AI model to serve as the foundation, so he had to train his own model based on his dad's data. Putting immense thought into the effort, Ahmad decided to close off the system from the rest of the Internet so that only his dad's memories would inform the model. To prevent unauthorized chats, he kept the bot on a laptop that only his family could access.

Ahmad was so intent on building a digital replica that felt just like his dad that it didn't occur to him until after his family started using the bot that he never asked his dad if this was what he wanted. Over time, he realized that the bot was biased to his view of his dad, perhaps even feeling off to his siblings who had a slightly different relationship with their father. It's unclear if his dad would similarly view the bot as preserving just one side of him.

Ultimately, Ahmad didn't regret building the bot, and he told Ars he thinks his father "would have been fine with it."

But he did regret not getting his father's consent.

For people creating bots today, seeking consent may be appropriate if there's any chance the bot may be publicly accessed, Ahmad suggested. He told Ars that he would never have been comfortable with the idea of his dad's digital replica being publicly available because the question of an "accurate representation" would come even more into play, as malicious actors could potentially access it and sully his dad's memory.

Today, anybody can use ChatGPT's model to freely create a similar bot with their own loved one's data. And a wide range of grief tech services have popped up online, including HereAfter AI, SeanceAI, and StoryFile, Axiosnotedin an October report detailing the latest ways "AI could be used to 'resurrect' loved ones." As this trend continues "evolving very fast," Ahmad told Ars that estate planning is probably the best way to communicate one's AI ghost preferences.

But in a recently publishedarticleon "The Law of Digital Resurrection," law professor Victoria Haneman warned that "there is no legal or regulatory landscape against which to estate plan to protect those who would avoid digital resurrection, and few privacy rights for the deceased. This is an intersection of death, technology, and privacy law that has remained relatively ignored until recently."

Haneman agreed with Sheehan that "existing protections are likely sufficient to protect against unauthorized commercial resurrections"—like when actors or musicians are resurrected for posthumous performances. However, she thinks that for personal uses, digital resurrections may best be blocked not through estate planning but by passing a "right to deletion" that would focus on granting the living or next of kin the rights to delete the data that could be used to create the AI ghost rather than regulating the output.

A "right to deletion" could help people fight inappropriate uses of their loved ones' data, whether AI is involved or not. After her article was published, a lawyer reached out to Haneman about a client's deceased grandmother whose likeness was used to create a meme of her dancing in a church. The grandmother wasn't a public figure, and the client had no idea "why or how somebody decided to resurrect her deceased grandmother," Haneman told Ars.

Although Haneman sympathized with the client, "if it's not being used for a commercial purpose, she really has no control over this use," Haneman said. "And she's deeply troubled by this."

Haneman's article offers a rare deep dive into the legal topic. It sensitively maps out the vague territory of digital rights of the dead and explains how those laws—or the lack thereof—interact with various laws dealing with death, from human remains to property rights.

In it, Haneman also points out that, on balance, the rights of the living typically outweigh the rights of the dead, and even specific instructions on how to handle human remains aren't generally considered binding. Some requests, like organ donation that can benefit the living, are considered critical, Haneman noted. But there are mixed results on how courts enforce other interests of the dead—like a famous writer's request to destroy all unpublished work or a pet lover's insistence to destroy their cat or dog at death.

She told Ars that right now, "a lot of people are like, 'Why do I care if somebody resurrects me after I'm dead?' You know, 'They can do what they want.' And they think that, until they find a family member who's been resurrected by a creepy ex-boyfriend or their dead grandmother's resurrected, and then it becomes a different story."

Existing law may protect "the privacy interests of the loved ones of the deceased from outrageous or harmful digital resurrections of the deceased," Haneman noted, but in the case of the dancing grandma, her meme may not be deemed harmful, no matter how much it troubles the grandchild to see her grandma's memory warped.

Limited legal protections may not matter so much if, culturally, communities end up developing a distaste for digital replicas, particularly if it becomes widely viewed as disrespectful to the dead, Haneman suggested. Right now, however, society is more fixated on solving other problems with deepfakes rather than clarifying the digital rights of the dead. That could be because few people have been impacted so far, or it could also reflect a broader cultural tendency to ignore death, Haneman told Ars.

"We don't want to think about our own death, so we really kind of brush aside whether or not we care about somebody else being digitally resurrected until it's in our face," Haneman said.

Over time, attitudes may change, especially if the so-called "digital afterlife industry" takes off. And there is some precedent that the law could be changed to reinforce any culture shift.

"The throughline revealed by the law of the dead is that a sacred trust exists between the living and the deceased, with an emphasis upon protecting common humanity, such that data afforded no legal status (or personal data of the deceased) may nonetheless be treated with dignity and receive some basic protections," Haneman wrote.

Preventing yourself from becoming an AI ghost seemingly now falls in a legal gray zone that policymakers may need to address.

Haneman calls for a solution that doesn't depend on estate planning, which she warned "is a structurally inequitable and anachronistic approach that maximizes social welfare only for those who do estate planning." More than 60 percent of Americans die without a will, often including "those without wealth," as well as women and racial minorities who "are less likely to die with a valid estate plan in effect," Haneman reported."We can do better in a technology-based world," Haneman wrote. "Any modern framework should recognize a lack of accessibility as an obstacle to fairness and protect the rights of the most vulnerable through approaches that do not depend upon hiring an attorney and executing an estate plan."

Rather than twist the law to "recognize postmortem privacy rights," Haneman advocates for a path for people resistant to digital replicas that focuses on a right to delete the data that would be used to create the AI ghost.

"Put simply, the deceased may exert control over digital legacy through the right to deletion of data but may not exert broader rights over non-commercial digital resurrection through estate planning," Haneman recommended.

Sheehan told Ars that a right to deletion would likely involve estate planners, too.

"If this is not addressed in an estate planning document and not specifically addressed in the statute (or deemed under the authority of the executor via statute), then the only way to address this would be to go to court," Sheehan said. "Even with a right of deletion, the deceased would need to delete said data before death or authorize his executor to do so post death, which would require an estate planning document, statutory authority, or court authority."

Haneman agreed that for many people, estate planners would still be involved, recommending that "the right to deletion would ideally, from the perspective of estate administration, provide for a term of deletion within 12 months." That "allows the living to manage grief and open administration of the estate before having to address data management issues," Haneman wrote, and perhaps adequately balances "the interests of society against the rights of the deceased."

To Haneman, it's also the better solution for the people left behind because "creating a right beyond data deletion to curtail unauthorized non-commercial digital resurrection creates unnecessary complexity that overreaches, as well as placing the interests of the deceased over those of the living."

If a dystopia that experts paint comes true, Big Tech companies may one day profit by targeting grieving individuals to seize the data of the dead, which could be more easily abused since it's granted fewer rights than data of the living.

Perhaps in that future, critics suggest, people will be tempted into free trials in moments when they're missing their loved ones most, then forced to either pay a subscription to continue accessing the bot or else perhaps be subjected to ad-based models where their chats with AI ghosts may even feature ads in the voices of the deceased.

Today, even in a world where AI ghosts aren't yet compelling ad clicks, some experts have warned that interacting with AI ghosts could cause mental health harms, New Scientistreported, especially if the digital afterlife industry isn't carefully designed, AI ethicistswarned. Some people may end up getting stuck maintaining an AI ghost if it's left behind as a gift, and ethicists suggested that the emotional weight of that could also eventually take a negative toll. While saying goodbye is hard, letting go is considered a critical part of healing during the mourning process, and AI ghosts may make that harder.

But the bots can be a helpful tool to manage grief, some experts suggest, provided that their use is limited to allow for a typical mourning process or combined with therapy from a trained professional, Al Jazeerareported. Ahmad told Ars that working on his bot has not only kept his father close to him but also helped him think more deeply about relationships and memory.

Haneman noted that people have many ways of honoring the dead. Some erect statues, and others listen to saved voicemails or watch old home movies. For some, just "smelling an old sweater" is a comfort. And creating digital replicas, as creepy as some people might find them, is not that far off from these traditions, Haneman said.

"Feeding text messages and emails into existing AI platforms such as ChatGPT and asking the AI to respond in the voice of the deceased is simply a change in degree, not in kind," Haneman said.

For Ahmad, the decision to create a digital replica of his dad was a learning experience, and perhaps his experience shows why any family or loved one weighing the option should carefully consider it before starting the process.

In particular, he warns families to be careful introducing young kids to grief bots, as they may not be able to grasp that the bot is not a real person. When he initially saw his young kids growing confused with whether their grandfather was alive or not—the introduction of the bot was complicated by the early stages of the pandemic, a time when they met many relatives virtually—he decided to restrict access to the bot until they were older. For a time, the bot only came out for special events like birthdays.

He also realized that introducing the bot also forced him to have conversations about life and death with his kids at ages younger than he remembered fully understanding those concepts in his own childhood.

Now, Ahmad's kids are among the first to be raised among AI ghosts. To continually enhance the family's experience, their father continuously updates his father's digital replica. Ahmad is currently most excited about recent audio advancements that make it easier to add a voice element. He hopes that within the next year, he might be able to use AI to finally nail down his South Asian father's accent, which up to now has always sounded "just off." For others working in this space, the next frontier is realistic video or even augmented reality tools, Ahmad told Ars.

To this day, the bot retains sentimental value for Ahmad, but, as Haneman suggested, the bot was not the only way he memorialized his dad. He also created a mosaic, and while his father never saw it, either, Ahmad thinks his dad would have approved.

"He would have been very happy," Ahmad said.

There's no way to predict how future generations may view grief tech. But while Ahmad said he's not sure he'd be interested in an augmented reality interaction with his dad's digital replica, kids raised seeing AI ghosts as a natural part of their lives may not be as hesitant to embrace or even build new features. Talking to Ars, Ahmad fondly remembered his young daughter once saw that he was feeling sad and came up with her own AI idea to help her dad feel better.

"It would be really nice if you can just take this program and we build a robot that looks like your dad, and then add it to the robot, and then you can go and hug the robot," she said, according to her father's memory.

Wondering what to get the dad who has everything? We have some ideas!

Greetings Arsians! It's time—at least in some parts of the world—to celebrate dads. Father's Day is nearly here, and as there's a custom of gift-giving, many of us will have to choose something. Below, various Ars editors have identified a few things they've bought recently that they think could be great gifts for dads—with the caveat that there's an indefinably large spectrum of variations of what dads like. Still, we did our best to include a few things that are pretty general, and a few that are weirdly specific. In any case, want to show some appreciation for your dad? Here are some options you can throw some money at.

Unless your father has light dimmers or has jumped onboard the small-lights-not-big-ones interior design trend, chances are there are two environments available to him for reading: one giant, bright ceiling light that is great for reading but not so great for setting an immersive and relaxing mood or, well, darkness. Enter the clip book light, a good way to light up the page at a brightness suitable for not-so-new eyes without harshing the reading vibes. The 86Lux Book Light is one of many offerings, but we like its simple design. It has multiple temperature and brightness levels, and it's adjustable, so it can reliably clip onto most books.

Stanley thermoses are renowned for their longevity and quality, with decades-old models still handling daily duty. TheStanley Heritage Thermosis a modern version of the classic containers, featuring a 1.1-quart capacity, stainless steel body, and the brand's trademark vacuum insulation that can keep dad's beverage of choice hot or cold for an impressive 24 hours. The lid is leak-proof, so you can toss it in a bag or backpack without worry, and it doubles as an 8-oz cup. At about $56, it's more spendy than your average bargain-bin plastic thermos, but you get what you pay for.

AVIDGRAM HDMI 2.1 Switch 4-Port- $90

OK, so this one is admittedly for a very specific dad—but he exists, and if he's your dad, you know it. Should your father be a true home theater geek and/or a console gamer, he has likely run up against the problem that most modern TVs just don't have enough HDMI ports. What's a dad to do when his TV has four HDMI ports, but he has a streaming box, a DVD player, a PlayStation, an Xbox, and a Switch? That's one too many devices. Enter the Avidgram HDMI 4-port switcher. It supports HDMI 2.1, so it works with modern game consoles. And of a handful we've tried, it's one of the nicest to use and most reliable. Plus, it has a remote, so dad doesn't have to get off the couch to switch devices.

LaCie Rugged USB-C, 4TB Portable External Hard Drive- $150

It’s always a good time to give the dad in your life peace of mind through data redundancy. That sounds boring, but it’s important. Whether dad is a digital packrat or just prudently paranoid about data loss, this rugged external drive offers a sizable 4TB of backup space at a price that makes copy-and-forget offsite storage feasible. The distinctive orange bumper isn't just for show—it protects against 4-foot drops, dust, and water splashes, making it ideal for stashing in a safe deposit box, glove compartment, or anywhere away from the original data source in case disaster strikes. USB-C ensures transfers happen at decent speeds (up to 5Gb/s), so backing up his photo library won't take all weekend. Mac users will need to reformat, but built-in password protection works regardless of platform.

The Apple TV 4K streaming box can be a suitable gift for various types of dads, from couch potatoes and tech fans to streaming enthusiasts and streaming holdouts. If your dad has been holding out on cutting the cord, the latest Apple TV box is a good incentive for him to make the move.

There’s a strong chance that the Apple TV’s tvOS operating system (OS) will work more smoothly and reliably than whatever OS his TV uses. If your dad has multiple streaming subscriptions, the set-top box’s built-in Apple TV app is handy for unifying many mainstream streaming libraries, so he can spend less time app-hopping and more time watching.

The Apple TV also offers one of themost private approaches to streaming. Your dad (or you) will find it easy and quick to set his privacy controls when setting up the Apple TV. And you don’t have to worry about the OS tracking your dad’s activity nearly as much as you do with most smart TVs and other streaming hardware.

For a bonus gift, you can sign your dad up for a relevant streaming subscription(s) that he doesn’t have, ensuring his new streaming box is put to use. You can save money here by adding an extra member to a streaming subscription you already have or installing free streaming apps.

The Sony WH-1000XM6 don't have the best name, but they are some of the nicest Bluetooth headphones money can buy. These cans offer superb audio fidelity, along with class-leading noise cancellation, which is great if dad is a jet-setter—or even if his household is just a bit chaotic. They also have a full raft of connectivity options, like Bluetooth LE, Auracast, and multipoint for pairing with multiple devices. The battery life is rated at an impressive 30 hours with noise cancellation on, and the frame and ear cups are generously padded for long listening sessions without discomfort. Plus, they can fold up for easier transport compared to the older XM5 headphones. The $450 price tag is higher than many other headphones, but it's hard to argue with the total package.

Segway Ninebot E2 Pro Scooter- $450

Fair warning: Dad will feel like a kid again on this thing—just insist on a high-quality helmet. The E2 Pro's 15.5 mph top speed is fast enough not to feel slow but slow enough not to be insanely dangerous, hitting that perfect sweet spot for urban commuting or a neighborhood joyride. With up to 25 miles of range, a dual braking system, and traction control for various terrains, it's practical transportation that happens to be fun. The 750 W peak motor handles 18 percent inclines, while Apple FindMy integration adds security. You can also lock it via an app on a smartphone. Safety gear is essential—this thing is more thrilling than the specs suggest—but that's exactly what makes it such a perfect gift.

iFixit FixHub Soldering Toolkit- $300

Today’s handyman needs more than just a tool belt and nails. Today, it’s very common for the things that need repairing to include some sort of circuit board or other electronic components. With the proper soldering kit, your dad could up his repair game and save more items from the trash bin. But not every dad was born with a soldering iron in their mouth.

iFixit built its FixHub Soldering Toolkit as an introductory and portable soldering iron that’s novice-friendly, yet remarkably helpful.

The full toolkit comes with a soldering iron, battery pack, and a variety of additional tools, including mandible wire strippers and flush cutters, silicone electrical tape, angled tweezers, a desoldering braid, and a silicone work mat.

If your dad doesn’t need all that, iFixit also sells the soldering iron with just the battery packfor $250.Or you could opt to buy the portable soldering iron with a USB-C charging cablefor $80.

In either case, your father gets a soldering iron with a swappable beveled 1.5 mm tip that powers up to 100 watts, besting the 60 W you typically see from USB-powered soldering irons. Your dad will be eagerly awaiting the next broken gadget.

Ars Technica may earn compensation for sales from links on this post throughaffiliate
programs.

Meta's Llama 4 model has underperformed its competitors on benchmarks.

Meta has invested $15 billion into data-labeling startup Scale AI and hired its co-founder, Alexandr Wang, as part of its bid to attract talent from rivals in a fiercely competitive market.

The deal values Scale at $29 billion, double its valuation last year. Scale said it would “substantially expand” its commercial relationship with Meta “to accelerate deployment of Scale’s data solutions,” without giving further details. Scale helps companies improve their artificial intelligence models by providing labeled training data.

Scale will distribute proceeds from Meta’s investment to shareholders, and Meta will own 49 percent of Scale’s equity following the transaction.

Wang, who was also Scale’s chief executive, will “work on Meta’s AI efforts” in his new role. Meta, however, did not reveal his new job title.

Wang, 28, is set to remain on the board of directors at Scale, which has named Jason Droege as chief strategy officer. Droege helped launch Uber Eats as interim CEO and joined Scale in September.

“Meta’s investment recognizes Scale’s accomplishments to date and reaffirms that our path forward—like that of AI—is limitless,” Wang said. “I’m delighted that Jason will lead the next steps in Scale’s journey.”

Meta’s investment is the latest attempt by chief executive Mark Zuckerberg to give his $1.8 trillion social media company an edge in the race to develop more powerful AI models. Zuckerberg has been trying to poach top researchers and engineers from rival groups as he seeks to build out a new “superintelligence” team.

The deal is one of the biggest of its kind as tech companies increasingly strike agreements to invest in startups while acquiring their top staff. Last year, Microsoft paid $650 million to hire Inflection boss Mustafa Suleyman and his top lieutenants, and to license the startup’s technology. Google also paid $2.7 billion for a similar arrangement with Character.AI.

Meta has invested heavily in generative AI, with the majority of its planned $72 billion in capital expenditure this year earmarked for data centers and servers. The deal underlines the high price AI companies are willing to pay for data that can be used to train AI models.

Zuckerberg pledged last year that his company’s models would outstrip rivals’ efforts in 2025, but Meta’s most recent release, Llama 4, has underperformed on various independent reasoning and coding benchmarks.

The long-term goal of researchers at Meta “has always been to reach human intelligence and go beyond it,” said Yann LeCun, the company’s chief AI scientist at the VivaTech conference in Paris this week.

Building artificial “general” intelligence—AI technologies that have human-level intelligence—is a popular goal for many AI companies. An increasing number of Silicon Valley groups are also seeking to reach “superintelligence,” a hypothetical scenario where AI systems surpass human intelligence.

The core of Scale’s business has been data-labeling, a manual process of ensuring images and text are accurately labeled and categorized before they are used to train AI models.

Wang has forged relationships with Silicon Valley’s biggest investors and technologists, including OpenAI’s Sam Altman. Scale AI’s early customers were autonomous vehicle companies, but the bulk of its expected $2 billion in revenues this year will come from labeling the data used to train the massive AI models built by OpenAI and others.

The deal will result in a substantial payday for Scale’s early venture capital investors, including Accel, Tiger Global Management, and Index Ventures. Tiger’s $200 million investment is worth more than $1 billion at the company’s new valuation, according to a person with knowledge of the matter.

Additional reporting by Tabby Kinder in San Francisco

© 2025 The Financial Times Ltd.All rights reserved.Not to be redistributed, copied, or modified in any way.

It was a good idea that arrived too late to matter.

Apps used to be the measure of a mobile platform's worth, with Apple and Google dueling over who could list the most items in their respective stores. Today, the numbers don't matter as much—there are enough apps, and Google's attempt to replace parts of the web with apps is going away.Instant Apps, a feature that debuted in 2017, will reportedly be scrapped in December 2025. In its place, you'll just have to use the Internet.

Developer Leon Omelan spotted this news buried in the latest Canary release of Android Studio (confirmed byAndroid Authority). The development client includes a warning that Instant Apps is headed for the Google graveyard. Here's the full notice, which is the only official confirmation from Google at this time.

Instant Apps wasn't a bad idea—it was just too late. Early in the mobile era, browsers and websites were sluggish on phones, making apps a much better option. Installing them for every site that offered them could be a pain, though. Google's Instant Apps tried to smooth over the experience by delivering an app live without installation. When developers implemented the feature, clicking a link to their websites could instead open the Android app in a similar amount of time as loading a webpage. Google laterexpanded the featureto games.

But that was then, and this is now. Today, an increasing number of mobile apps are functionally identical to the mobile websites they are intended to replace, and developer uptake of Instant Apps was minimal. Even in 2017, loading an app instead of a website had limited utility. As a result, most of us probably only encountered Instant Apps a handful of times in all the years it was an option for developers.

To use the feature, which was delivered to virtually all Android devices by Google Play Services, developers had to create a special "instant" version of their app that was under 15MB. The additional legwork to get an app in front of a subset of new users meant this was always going to be a steep climb, and Google struggles to incentivize developers to adopt new features. Plus, there's no way to cram in generative AI! So it's not a shock to see Google retiring the feature.

This feature is currently listed in the collection of Google services in your phone settings as "Google Play Instant." Unfortunately, there aren't many examples still available if you're curious about what Instant Apps were like—the Finnish publisherIlta-Sanomatis one of the few still offering it. Make sure the settings toggle for Instant Apps is on if you want a little dose of nostalgia.

Nonprofits often can't afford ad space on streaming platforms.

DENVER—Ads shown while you're streaming shows or movies are disruptive enough. But there's something uniquely eerie about what you see when a connected TV (CTV) platform fails to sell ad inventory. You may get a black screen accompanied by ethereal music or a confusing thumping beat, alongside a graphic that says something like, "We'll be right back."

Not only are streamers being forced to enduremore ad time than ever,but that time doesn't even always benefit streaming platforms or advertisers. For the past six months, AdGood has been working to turn that blank, wasted ad space into messaging for good by helping nonprofits buy ad space for cheap.

During the StreamTV Show in Denver this week, Ars spoke with Kris Johns, CEO and founder ofAdGood, a 501(c)(3) nonprofit that sells unused, CTV ad space to other nonprofits. AdGood sells unfilled, sometimes donated, ad space at discounted rates, whichit sayscan be as low as about $5 to $6 CPMs (cost per mille, or the amount an advertiser pays for every 1,000 impressions an ad earns). Johns said that CTV CPMs can vary depending on the scenario, with costs ranging from $12 to $15 and higher. Some CTV ad firmspegthe average CTV CPM at$35 to $65.

According to Johns, AdGood currently has “about 60-plus” nonprofit customers across the US that span different sizes and verticals. Its customer base covers various areas, including animals, education, health care, and homelessness. Johns said AdGood would like to be in all vertical sectors eventually.

AdGood is able to sell nonprofits CTV ad space for cheap through partnerships, including ones with LG Ad Solutions and, announced this week, Stingray and Plex.

"Partnering with AdGood allows us to extend the impact of our media platform to nonprofits doing critical work around the world,” Shawn Eldridge, VP of business development and Content at Plex, said in a statement.

AdGood’s offerings also include a managed service for ad campaign management for nonprofits. AdGood doesn’t yet offer pixels, but Johns said developments like that are “in the works.”

Johns explained that while many nonprofits use services like Meta and Google AdWords for tracking ads, they’re “hitting plateaus” with their typical methods. He said there is nonprofit interest in reaching younger audiences, who often use CTV devices:

A lot of them have been looking for ways to get [into CTV ads], but, unfortunately, with minimum spend amounts, they're just not able to access it.

AdGood also sells a self-serve generative AI ad manager, which it offers via a partnership withStreamr.AI. The tool is designed to simplify the process of creating 30-second video ads that are “completely editable via a chat prompt,” according to Johns.

“It automatically generates all their targeting. They can update their targeting for whatever they want, and then they can swipe a credit card and essentially run that campaign. It goes into our approval queue, which typically takes 24 hours for us to approve because it needs to be deemed TV-quality," he explained.

The executive said AdGood charges nonprofits a $7 CPM and a $250 flat fee for the service. He added:

Think about a small nonprofit in a local community, for instance, my son's special needs baseball team. I can get together with five other parents, easily pull together a campaign, and run it in our local town. We get seven kids to show up, and it changes their lives. We’re talking about $250 having a massive impact in a local market.

Looking ahead, Johns said he’d like to see AdGood’s platform and team grow to be able to give every customer “a certain allocation of inventory, whether it's 50,000 impressions a month or 100,000 a month.”

For some, streaming ads are rarely a good thing. But when those ads can help important causesandreplace odd blank ad spaces that make us question our own existence, it brings new meaning to the idea of a "good" commercial.

A crew mission has been delayed until next Wednesday at the earliest.

There's another leak on the International Space Station, and NASA has already delayed one crew launch to the orbiting laboratory as a result.

Beyond that, the space agency is not offering much information about the unfolding situation in orbit. However, multiple sources have confirmed to Ars that the leak is a serious concern for the space agency as it deals with hardware that is approaching three decades in orbit.

To understand the current situation, it is important to review past leaks on the station, which has an aluminum structure. The station has had a slowbut increasing leak since 2019. The air leaks are located in the transfer tunnel of the space station's Russian Zvezda service module, one of the oldest elements of the complex, the first elements of which were launched in 1998. The transfer tunnel, known by the Russian acronym PrK, connects the Zvezda module with a docking port where Soyuz crew and Progress resupply spacecraft attach to the station.

From time to time, Russian cosmonauts have experimented with repairs to the small cracks, but they have generally only slowed the progression of the leak, which amounts to a couple of pounds of air per day. The best solution has been to close the hatch leading to the PrK module except when spacecraft dock with the attached port.

Roscosmosrecently confirmedthat it completed its most recent repairs on the PrK module earlier this month, saying it had been "completely sealed." NASA has also said that repairs were recently wrapped up. After this, both Roscosmos and NASA said the leak rate inside the PrK module had halted.

This seems like good news. However, the overall air pressure in the space station at large continued to drop, according to two sources. So if the PrK module was not leaking, as it had been doing for half a decade, why was the space station still losing air pressure?

No one is certain. The best guess is that the seals on the hatch leading to the PrK module are, in some way, leaking. In this scenario, pressure from the station is feeding the leak inside the PrK module through these seals, leading to a stable pressure inside—making it appear as though the PrK module leaks are fully repaired.

At this point, NASA is monitoring the ongoing leak and preparing for any possibility. A senior industry source told Ars that the NASA leadership of the space station program is "worried" about the leak and its implications.

This is one reason the space agency delayed the launch of a commercial mission carrying four astronauts to the space station, Axiom-4, on Thursday.

"The postponement of Axiom Mission 4 provides additional time for NASA and Roscosmos to evaluate the situation and determine whether any additional troubleshooting is necessary," NASAsaid in a statement. "A new launch date for the fourth private astronaut mission will be provided once available."

One source indicated that the new tentative launch date is now June 18. However, this will depend on whatever resolution there is to the leak issue.

The worst-case scenario for the space station is that the ongoing leaks are a harbinger of a phenomenon known as "high cycle fatigue," which affects metal, including aluminum. Consider that if you bend a metal clothes hanger once, it bends. But if you bend it back and forth multiple times, it will snap. This is because, as the metal fatigues, it hardens and eventually snaps. This happens suddenly and without warning, as was the case with anAloha Airlines flight in 1988.

The concern is that some of these metal structures on board the station could fail quickly and catastrophically. Accordingly, in its previous assessments,NASA has classifiedthe structural cracking issue on the space station as the highest level of concern on its 5v5 risk matrix to gauge the likelihood and severity of risks to the space station.

In the meantime, the space agency has not been forthcoming with any additional information. Despite many questions from Ars Technica and other publications, NASA has not scheduled a press conference or said anything else publicly about the leaks beyond stating, "The crew aboard the International Space Station is safely conducting normal operations."

Report: Proposed condition on ad merger would ban boycotts based on politics.

The Federal Trade Commission is reportedly pitching a merger condition that would forbid advertising agencies from boycotting platforms based on political content, in a move that could benefit Elon Musk's X social network and President Trump's own Truth Social platform.

As the FTC reviews a proposed merger between Omnicom Group and Interpublic Group, two large ad agencies, The New York Timesreportedyesterday that a "proposed consent decree would prevent the merged company from boycotting platforms because of their political content by refusing to place their clients' advertisements on them, according to two people briefed on the matter."

This is one of several moves the FTC has reportedly made to discourage ad boycotts that have riled conservatives. The FTC currently has only Republican commissioners because President Trumpfired both Democrats, who allege in alawsuitthat the firings were illegal. Trump alsodeclared sweeping executive powerover the FTC and other agencies that were created to operate independently from the White House.

On Monday, the FTC sent civil investigative demands to Omnicom, Interpublic, and several other large ad agencies "as part of an investigation into whether advertising and advocacy groups violated antitrust laws by coordinating boycotts of certain sites, including Elon Musk's X," The Wall Street Journalreported.

The FTC also "demanded documents from Media Matters about possible coordination with other media watchdogs accused by Elon Musk of helping orchestrate advertiser boycotts of X," Reutersreportedon May 22. X has apending lawsuitagainst Media Matters andanother againstthe World Federation of Advertisers (WFA) and several large corporations. Law professors have said thatadvertisers have a strong defense under the First Amendment, but Musk's firm reportedly cajoled some companies into buying ads bythreateningto sue them.

After Musk's purchase of Twitter, the social network lost advertisers for various reasons, including changes to content moderation and an incident in which Muskposted a favorable responseto an antisemitic tweet and then told concerned advertisers to "go fuck yourself."

FTC Chairman Andrew Ferguson said at aconference in Aprilthat "the risk of an advertiser boycott is a pretty serious risk to the free exchange of ideas."

"If advertisers get into a back room and agree, 'We aren't going to put our stuff next to this guy or woman or his or her ideas,' that is a form of concerted refusal to deal," Ferguson said. "The antitrust laws condemn concerted refusals to deal. Now, of course, because of the First Amendment, we don't have a categorical antitrust prohibition on boycotts. When a boycott ceases to be economic for purposes of the antitrust laws and becomes purely First Amendment activity, the courts have not been super clear—[it's] sort of a 'we know it when we see it' type of thing."

The FTCwebsitesays that any individual company acting on its own may "refuse to do business with another firm, but an agreement among competitors not to do business with targeted individuals or businesses may be an illegal boycott, especially if the group of competitors working together has market power." The examples given on the FTC webpage are mostly about price competition and do not address the widespread practice of companies choosing where to place advertising based on concerns about their brands.

We contacted the FTC about the merger review today and will update this article if it provides any comment.

X'slawsuittargets a World Federation of Advertisers initiative called the Global Alliance for Responsible Media (GARM), a now-defunct program that Omnicom and Interpublic participated in. X itself waspart of the GARM initiative, whichshut downafter X filed the lawsuit. X alleged that the defendants conspired "to collectively withhold billions of dollars in advertising revenue."

The World Federation of Advertisers said in acourt filinglast month that GARM was founded "to bring clarity and transparency to disparate definitions and understandings in advertising and brand safety in the context of social media. For example, certain advertisers did not want platforms to advertise their brands alongside content that could negatively impact their brands."

GARM's goal was to define what counts as "violent" or "obscene" content, because "these concepts are unclear and extraordinarily subjective… GARM sought to end this confusion and create a transparent, brand-safety framework free from the ambiguities of what, for example, advertisers meant when using the term 'violence.' In short, GARM created a common 'language' that the relevant industry could use," the filing said.

The World Federation of Advertisers, which is a Belgian nonprofit, is trying to get X's lawsuit dismissed by arguing that the US District Court for the Northern District of Texas is an improper venue for the lawsuit.

Senate Democrats have raised concerns about Musk potentially using his position as a special government employee to block the Omnicom/Interpublic merger unless advertisers spend more on X.

"There has beenreportingthat Interpublic received a call from a lawyer at X threatening that if Interpublic did not get its clients to spend more on ads on Musk's social media platform there would be consequence," five Senate Democrats wrote in a Marchletterto Ferguson. "This reporting states that X's CEO, Linda Yaccarino, made similar warnings in conversations with executives of the advertising company. Interpublic has reportedly interpreted these communications to mean that Musk will leverage his influence over President Trump to stall or block Interpublic's $13 billion deal to merge with advertising competitor Omnicom Group."

X could "be attempting to strike a quid-pro-quo deal, pressuring Interpublic to get its clients to spend a certain amount on advertising on X in exchange for directing President Trump to use his antitrust enforcement agencies to allow Interpublic's merger with Omnicom to proceed," the letter said. The Democrats sent a similar letter to Attorney General Pam Bondi. Musk's time as a special employee in the Trump administrationendedin late May.

Because you wanted this, right?

NotebookLM is undoubtedly one of Google's best implementations of generative AI technology, giving you the ability to explore documents and notes with a Gemini AI model. Last year, Google added the ability to generate so-called "audio overviews" of your source material in NotebookLM. Now, Google has brought thosefake AI podcaststo search results as a test. Instead of clicking links or reading the AI Overview, you can have two nonexistent peopletell you what the results say.

This feature is not currently rolling out widely—it's available in search labs, which means you have tomanually enable it. Anyone can opt in to the new Audio Overview search experience, though. If you join the test, you'll quickly see the embedded player in Google search results. However, it's not at the top with the usual block of AI-generated text. Instead, you'll see it after the first few search results, below the "People also ask" knowledge graph section.

Credit:

Google

Google isn't wasting resources to generate the audio automatically, so you have to click the generate button to get started. A few seconds later, you're given a back-and-forth conversation between two AI voices summarizing the search results. The player includes a list of sources from which the overview is built, as well as the option to speed up or slow down playback.

Google suggests searching for "how do noise cancellation headphones work" as an example, but we're seeing Audio Overviews appear in a lot of searches. You can even search for "Google audio overviews" to hear an Audio Overview explain what an Audio Overview is. That's neat in a recursive technological horror kind of way.

The overviews we've listened to are mostly accurate for simple topics, but there are many more opportunities for errors here. In NotebookLM, the AI is grounded by the small collection of documents and websites you feed into the system. With search results, there are more opportunities for the AI to go down the wrong path while summarizing. This is a problem we'veseen repeatedlywith text AI Overviews.

Search is not the first expansion of Google's fake podcasts. Since debuting in NotebookLM, Audio Overviews have also shown up in Google's Gemini Deep Research mode and Google Docs. Given Google's rapid advancement of AI features, we expect to see Audio Overviews in even more places going forward. And don't be surprised if Audio Overviews invade your search results soon, even if you skip this test. After all, the text-based AI Overviews were only in labs for a couple of months before they graduated to the default search experience.

Combining vertical "strings" with periodic horizontal wires stops clogging and clumping, boosts efficiency.

Arid coastal regions that are also prone to fog are prime locations for fog-harvesting devices as a water source, especially during prolonged droughts. But the standard technology is prone to clogging. Scientists at Virginia Tech have created an improved version of their earlier "fog harp" alternative design to address that issue, according to anew paperpublished in the Journal of Materials Chemistry A.

Fog harvesting (or dew catching) is an ancient practice dating as far back as the Incas, who placed buckets under trees to collect condensation. It's also practiced by certain insects, notably Namib desert beetles, which survive on the water that condenses onto their wings. The wings have alternating hydrophilic and hydrophobic regions to enhance the condensation. Today's fog harvesters are usually mesh nets mounted onto frames with a trough or basin underneath. Like the beetle's wings, the mesh filaments are chemically coated to be either hydrophobic or hydrophilic.

The efficiency of these water harvesters depends on decreasing the size of the filaments and the mesh holes. "If the holes were too big, the microscopic droplets would pass through it, and it wouldn't harvest much water," co-author James Kaindu, a student in researcher Jonathan Boreyko's lab at Virginia Tech, told Ars. The trade-off is that smaller filaments and holes are more prone to clogging. "If it was too small, the droplets would coalesce and create a water film on it," said Kaindu. "It would impede the flow and act as a barrier that would dramatically affect its capture efficiency."

That's why Boreyko's group came up withthe fog harpin 2018. As the name implies, the fog harp consists of a frame holding vertical wires with no criss-crossing horizontal ones. It was inspired by California redwoods, which derive as much as 35 percent of their annual water intake from fog dripping from the parallel arrays of the trees' needle-like leaves.

The design performed very well in the laboratory during scale model testing, collecting between two to seven times the amount of water as mesh nets. But then they tested their design in Monterey Bay. That real-world deployment required larger versions on the order of 1 meter by 1 meter (or 3×3 feet). At that larger scale, surface tension becomes an issue, and the vertical strands start to clump together, "like your hair when it gets clumped up with water," said Kaindu. And once again, there are reduced harvesting efficiencies.

So the team decided upon a happy medium, combining elements of the traditional mesh with the fog harp to create a hybrid version: essentially a series of small-scale model fog harps stacked on top of each other, separated by the occasional cross-support wire. Boreyko likens the structure to a guitar neck with clearly demarcated frets. This eliminated the clumping problem to get the desired water harvesting efficiency.

Per Kaindu, the team tested variations of the concept in the lab, from standard mesh nets on one end, with their many interconnects, up to the original fog harp on the other extreme, with none. In between were versions with a hundred interconnects, 10 interconnects, five interconnects, and three interconnects. All were suspended in the air and blasted with fog from ultrasonic humidifiers.

Next they measured the mass of collected water after a specified time period to arrive at the overall efficiency—the fraction of all the fog water in the air relative to how much was captured—for each variation. The improved hybrid fog harp increased fog harvesting efficiency by a factor of between two and eight compared to pure mesh fog harvesters and the original fog harp, depending on other variables. The designs with three and five interconnects proved to be the sweet spot for the highest efficiency.

Unlike standard fog harvesting technologies, "We're trying to use clever geometric designs in place of chemistry," Boreyko told Ars. "When I first came into this field, virtually everyone was using nets, but they were just trying to make more and more clever chemical coatings to put on the nets to try to reduce the clogging. We found that simply going from a net to a harp, with no chemicals or coatings whatsoever—just the change in geometry solved the clogging problem much better."

For their scale prototypes in the lab, Boreyko's team 3D printed their harp "strings" out of a weakly hydrophobic plastic. "But in general, the harp works fantastic with uncoated stainless steel wires and definitely doesn't require any kind of fancy coating," said Boreyko. And the hybrid harp can be scaled up with relative ease, just like classic nets. It just means stringing together a bunch of harps of smaller heights, meter by meter, to get the desired size. "There is no limit to how big this thing could be," he said.

Scaling up the model is the next obvious step, along with testing larger prototypes outdoors. Boreyko would also like to test an electric version of the hybrid fog harp. "If you apply a voltage, it turns out you can catch even more water," he said. "Because our hybrid's non-clogging, you can have the best of both worlds: using an electric field to boost the harvesting amount in real-life systems and at the same time preventing clogging."

While the hybrid fog harp is well-suited for harvesting water in any coastal region that receives a lot of fog, Boreyko also envisions other, less obvious potential applications for high-efficiency fog harvesters, such as roadways, highways, or airport landing strips that are prone to fog that can pose safety hazards. "There's even industrial chemical supply manufacturers creating things like pressurized nitrogen gas," he said. "The process cools the surrounding air into an ice fog that can drift across the street and wreak havoc on city blocks."

Journal of Materials Chemistry A, 2025. DOI:10.1039/d5ta02686e(About DOIs).