Author name: Mike M.

trump-caving-on-nvidia-h20-export-curbs-may-disrupt-his-bigger-trade-war

Trump caving on Nvidia H20 export curbs may disrupt his bigger trade war

But experts seem to fear that Trump isn’t paying enough attention to how exports of US technology could threaten to not only supercharge China’s military and AI capabilities but also drain supplies that US firms need to keep the US at the forefront of AI innovation.

“More chips for China means fewer chips for the US,” experts said, noting that “China’s biggest tech firms, including Tencent, ByteDance, and Alibaba,” have spent $16 billion on bulk-ordered H20 chips over the past year.

Meanwhile, “projected data center demand from the US power market would require 90 percent of global chip supply through 2030, an unlikely scenario even without China joining the rush to buy advanced AI chips,” experts said. If Trump doesn’t intervene, one of America’s biggest AI rivals could even end up driving up costs of AI chips for US firms, they warned.

“We urge you to reverse course,” the letter concluded. “This is not a question of trade. It is a question of national security.”

Trump says he never heard of Nvidia before

Perhaps the bigger problem for Trump, national security experts suggest, would be if China or other trade partners perceive the US resolve to wield export controls as a foreign policy tool to be “weakened” by Trump reversing course on H20 controls.

They suggested that Trump caving on H20 controls could even “embolden China to seek additional access concessions” at a time when some analysts suggest that China may already have an upper hand in trade negotiations.

The US and China are largely expected to extend a 90-day truce following recent talks in Stockholm, Reuters reported. Anonymous sources told the South China Morning Post that the US may have already agreed to not impose any new tariffs or otherwise ratchet up the trade war during that truce, but that remains unconfirmed, as Trump continues to warn that chip tariffs are coming soon.

Trump has recently claimed that he thinks he may be close to cementing a deal with China, but it appears likely that talks will continue well into the fall. A meeting between Trump and Chinese President Xi Jinping probably won’t be scheduled until late October or early November, Reuters reported.

Trump caving on Nvidia H20 export curbs may disrupt his bigger trade war Read More »

openai’s-chatgpt-agent-casually-clicks-through-“i-am-not-a-robot”-verification-test

OpenAI’s ChatGPT Agent casually clicks through “I am not a robot” verification test

The CAPTCHA arms race

While the agent didn’t face an actual CAPTCHA puzzle with images in this case, successfully passing Cloudflare’s behavioral screening that determines whether to present such challenges demonstrates sophisticated browser automation.

To understand the significance of this capability, it’s important to know that CAPTCHA systems have served as a security measure on the web for decades. Computer researchers invented the technique in the 1990s to screen bots from entering information into websites, originally using images with letters and numbers written in wiggly fonts, often obscured with lines or noise to foil computer vision algorithms. The assumption is that the task will be easy for humans but difficult for machines.

Cloudflare’s screening system, called Turnstile, often precedes actual CAPTCHA challenges and represents one of the most widely deployed bot-detection methods today. The checkbox analyzes multiple signals, including mouse movements, click timing, browser fingerprints, IP reputation, and JavaScript execution patterns to determine if the user exhibits human-like behavior. If these checks pass, users proceed without seeing a CAPTCHA puzzle. If the system detects suspicious patterns, it escalates to visual challenges.

The ability for an AI model to defeat a CAPTCHA isn’t entirely new (although having one narrate the process feels fairly novel). AI tools have been able to defeat certain CAPTCHAs for a while, which has led to an arms race between those that create them and those that defeat them. OpenAI’s Operator, an experimental web-browsing AI agent launched in January, faced difficulty clicking through some CAPTCHAs (and was also trained to stop and ask a human to complete them), but the latest ChatGPT Agent tool has seen a much wider release.

It’s tempting to say that the ability of AI agents to pass these tests puts the future effectiveness of CAPTCHAs into question, but for as long as there have been CAPTCHAs, there have been bots that could later defeat them. As a result, recent CAPTCHAs have become more of a way to slow down bot attacks or make them more expensive rather than a way to defeat them entirely. Some malefactors even hire out farms of humans to defeat them in bulk.

OpenAI’s ChatGPT Agent casually clicks through “I am not a robot” verification test Read More »

meta-pirated-and-seeded-porn-for-years-to-train-ai,-lawsuit-says

Meta pirated and seeded porn for years to train AI, lawsuit says

Evidence may prove Meta seeded more content

Seeking evidence to back its own copyright infringement claims, Strike 3 Holdings searched “its archive of recorded infringement captured by its VXN Scan and Cross Reference tools” and found 47 “IP addresses identified as owned by Facebook infringing its copyright protected Works.”

The data allegedly demonstrates a “continued unauthorized distribution” over “several years.” And Meta allegedly did not stop its seeding after Strike 3 Holdings confronted the tech giant with this evidence—despite the IP data supposedly being verified through an industry-leading provider called Maxmind.

Strike 3 Holdings shared a screenshot of MaxMind’s findings. Credit: via Strike 3 Holdings’ complaint

Meta also allegedly attempted to “conceal its BitTorrent activities” through “six Virtual Private Clouds” that formed a “stealth network” of “hidden IP addresses,” the lawsuit alleged, which seemingly implicated a “major third-party data center provider” as a partner in Meta’s piracy.

An analysis of these IP addresses allegedly found “data patterns that matched infringement patterns seen on Meta’s corporate IP Addresses” and included “evidence of other activity on the BitTorrent network including ebooks, movies, television shows, music, and software.” The seemingly non-human patterns documented on both sets of IP addresses suggest the data was for AI training and not for personal use, Strike 3 Holdings alleged.

Perhaps most shockingly, considering that a Meta employee joked “torrenting from a corporate laptop doesn’t feel right,” Strike 3 Holdings further alleged that it found “at least one residential IP address of a Meta employee” infringing its copyrighted works. That suggests Meta may have directed an employee to torrent pirated data outside the office to obscure the data trail.

The adult site operator did not identify the employee or the major data center discussed in its complaint, noting in a subsequent filing that it recognized the risks to Meta’s business and its employees’ privacy of sharing sensitive information.

In total, the company alleged that evidence shows “well over 100,000 unauthorized distribution transactions” linked to Meta’s corporate IPs. Strike 3 Holdings is hoping the evidence will lead a jury to find Meta liable for direct copyright infringement or charge Meta with secondary and vicarious copyright infringement if the jury finds that Meta successfully distanced itself by using the third-party data center or an employee’s home IP address.

“Meta has the right and ability to supervise and/or control its own corporate IP addresses, as well as the IP addresses hosted in off-infra data centers, and the acts of its employees and agents infringing Plaintiffs’ Works through their residential IPs by using Meta’s AI script to obtain content through BitTorrent,” the complaint said.

Meta pirated and seeded porn for years to train AI, lawsuit says Read More »

ars-spoke-with-the-military’s-chief-orbital-traffic-cop—here’s-what-we-learned

Ars spoke with the military’s chief orbital traffic cop—here’s what we learned


“We have some 2,000 or 2,200 objects that I call the ‘red order of battle.'”

Col. Raj Agrawal participates in a change of command ceremony to mark his departure from Mission Delta 2 at Peterson Space Force Base, Colorado. Col. Barry Croker became the new commander of Mission Delta 2 on July 3.

For two years, Col. Raj Agrawal commanded the US military unit responsible for tracking nearly 50,000 human-made objects whipping through space. In this role, he was keeper of the orbital catalog and led teams tasked with discerning whether other countries’ satellites, mainly China and Russia, are peaceful or present a military threat to US forces.

This job is becoming more important as the Space Force prepares for the possibility of orbital warfare.

Ars visited with Agrawal in the final weeks of his two-year tour of duty as commander of Mission Delta 2, a military unit at Peterson Space Force Base, Colorado. Mission Delta 2 collects and fuses data from a network of sensors “to identify, characterize, and exploit opportunities and mitigate vulnerabilities” in orbit, according to a Space Force fact sheet.

This involves operating radars and telescopes, analyzing intelligence information, and “mapping the geocentric space terrain” to “deliver a combat-ready common operational picture” to military commanders. Agrawal’s job has long existed in one form or another, but the job description is different today. Instead of just keeping up with where things are in space—a job challenging enough—military officials now wrestle with distinguishing which objects might have a nefarious purpose.

From teacher to commander

Agrawal’s time at Mission Delta 2 ended on July 3. His next assignment will be as Space Force chair at the National Defense University. This marks a return to education for Agrawal, who served as a Texas schoolteacher for eight years before receiving his commission as an Air Force officer in 2001.

“Teaching is, I think, at the heart of everything I do,” Agrawal said. 

He taught music and math at Trimble Technical High School, an inner city vocational school in Fort Worth. “Most of my students were in broken homes and unfortunate circumstances,” Agrawal said. “I went to church with those kids and those families, and a lot of times, I was the one bringing them home and taking them to school. What was [satisfying] about that was a lot of those students ended up living very fulfilling lives.”

Agrawal felt a calling for higher service and signed up to join the Air Force. Given his background in music, he initially auditioned for and was accepted into the Air Force Band. But someone urged him to apply for Officer Candidate School, and Agrawal got in. “I ended up on a very different path.”

Agrawal was initially accepted into the ICBM career field, but that changed after the September 11 attacks. “That was a time with anyone with a name like mine had a hard time,” he said. “It took a little bit of time to get my security clearance.”

Instead, the Air Force assigned him to work in space operations. Agrawal quickly became an instructor in space situational awareness, did a tour at the National Reconnaissance Office, then found himself working at the Pentagon in 2019 as the Defense Department prepared to set up the Space Force as a new military service. Agrawal was tasked with leading a team of 100 people to draft the first Space Force budget.

Then, he received the call to report to Peterson Space Force Base to take command of what is now Mission Delta 2, the inheritor of decades of Air Force experience cataloging everything in orbit down to the size of a softball. The catalog was stable and predictable, lingering below 10,000 trackable objects until 2007. That’s when China tested an anti-satellite missile, shattering an old Chinese spacecraft into more than 3,500 pieces large enough to be routinely detected by the US military’s Space Surveillance Network.

This graph from the European Space Agency shows the growing number of trackable objects in orbit. Credit: European Space Agency

Two years later, an Iridium communications satellite collided with a defunct Russian spacecraft, adding thousands more debris fragments to low-Earth orbit. A rapid uptick in the pace of launches since then has added to the problem, further congesting busy orbital traffic lanes a hundred miles above the Earth. Today, the orbital catalog numbers roughly 48,000 objects.

“This compiled data, known as the space catalog, is distributed across the military, intelligence community, commercial space entities, and to the public, free of charge,” officials wrote in a fact sheet describing Mission Delta 2’s role at Space Operations Command. Deltas are Space Force military units roughly equivalent to a wing or group command in the Air Force.

The room where it happens

The good news is that the US military is getting better at tracking things in space. A network of modern radars and telescopes on the ground and in space can now spot objects as small as a golf ball. Space is big, but these objects routinely pass close to one another. At speeds of nearly 5 miles per second, an impact will be catastrophic.

But there’s a new problem. Today, the US military must not only screen for accidental collisions but also guard against an attack on US satellites in orbit. Space is militarized, a fact illustrated by growing fleets of satellites—primarily American, Chinese, and Russian—capable of approaching another country’s assets in orbit, and in some cases, disable or destroy them. This has raised fears at the Pentagon that an adversary could take out US satellites critical for missile warning, navigation, and communications, with severe consequences impacting military operations and daily civilian life.

This new reality compelled the creation of the Space Force in 2019, beginning a yearslong process of migrating existing Air Force units into the new service. Now, the Pentagon is posturing for orbital warfare by investing in new technologies and reorganizing the military’s command structure.

Today, the Space Force is responsible for predicting when objects in orbit will come close to one another. This is called a conjunction in the parlance of orbital mechanics. The US military routinely issues conjunction warnings to commercial and foreign satellite operators to give them an opportunity to move their satellites out of harm’s way. These notices also go to NASA if there’s a chance of a close call with the International Space Station (ISS).

The first Trump administration approved a new policy to transfer responsibility for these collision warnings to the Department of Commerce, allowing the military to focus on national security objectives.

But the White House’s budget request for next year would cancel the Commerce Department’s initiative to take over collision warnings. Our discussion with Agrawal occurred before the details of the White House budget were made public last month, and his comments reflect official Space Force policy at the time of the interview. “In uniform, we align to policy,” Agrawal wrote on his LinkedIn account. “We inform policy decisions, but once they’re made, we align our support accordingly.”

US Space Force officials show the 18th Space Defense Squadron’s operations floor to officials from the German Space Situational Awareness Centre during an “Operator Exchange” event at Vandenberg Space Force Base, California, on April 7, 2022. Credit: US Space Force/Tech. Sgt. Luke Kitterman

Since our interview, analysts have also noticed an uptick in interesting Russian activity in space and tracked a suspected Chinese satellite refueling mission in geosynchronous orbit.

Let’s rewind the tape to 2007, the time of China’s game-changing anti-satellite test. Gen. Chance Saltzman, today the Space Force’s Chief of Space Operations, was a lieutenant colonel in command of the Air Force’s 614th Space Operations Squadron at the time. He was on duty when Air Force operators first realized China had tested an anti-satellite missile. Saltzman has called the moment a “pivot point” in space operations. “For those of us that are neck-deep in the business, we did have to think differently from that day on,” Saltzman said in 2023.

Agrawal was in the room, too. “I was on the crew that needed to count the pieces,” he told Ars. “I didn’t know the significance of what was happening until after many years, but the Chinese had clearly changed the nature of the space environment.”

The 2007 anti-satellite test also clearly changed the trajectory of Agrawal’s career. We present part of our discussion with Agrawal below, and we’ll share the rest of the conversation tomorrow. The text has been lightly edited for brevity and clarity.

Ars: The Space Force’s role in monitoring activities in space has changed a lot in the last few years. Can you tell me about these changes, and what’s the difference between what you used to call Space Situational Awareness, and what is now called Space Domain Awareness?

Agrawal: We just finished our fifth year as a Space Force, so as a result of standing up a military service focused on space, we shifted our activities to focus on what the joint force requires for combat space power. We’ve been doing space operations for going on seven decades. I think a lot of folks think that it was a rebranding, as opposed to a different focus for space operations, and it couldn’t be further from the truth. Compared to Space Domain Awareness (SDA), Space Situational Awareness (SSA) is kind of the knowledge we produce with all these sensors, and anybody can do space situational awareness. You have academia doing that. You’ve got commercial, international partners, and so on. But Space Domain Awareness, Gen. [John “Jay”] Raymond coined the term a couple years before we stood up the Space Force, and he was trying to get after, how do we create a domain focused on operational outcomes? That’s all we could say at the time. We couldn’t say war-fighting domain at the time because of the way of our policy, but our policy shifted to being able to talk about space as a place where, not that we want to wage war, but that we can achieve objectives, and do that with military objectives in mind.

We used to talk about detect, characterize, attribute, predict. And then Gen. [Chance] Saltzman added target onto the construct for Space Domain Awareness, so that we’re very much in the conversation of what it means to do a space-enabled attack and being able to achieve objectives in, from, and to space, and using Space Domain Awareness as a vehicle to do those things. So, with Mission Delta 2, what he did is he took the sustainment part of acquisition, software development, cyber defense, intelligence related to Space Domain Awareness, and then all the things that we were doing in Space Domain Awareness already, put all that together under one command … and called us Mission Delta 2. So, the 18th Space Defense Squadron … that used to kind of be the center of the world for Space Domain Awareness, maybe the only unit that you could say was really doing SDA, where everyone else was kind of doing SSA. When I came into command a couple years ago, and we face now a real threat to having space superiority in the space domain, I disaggregated what we were doing just in the 18th and spread out through a couple of other units … So, that way everyone’s got kind of majors and minors, but we can quickly move a mission in case we get tested in terms of cyber defense or other kinds of vulnerabilities.

This multi-exposure image depicts a satellite-filled sky over Alberta. Credit: Alan Dyer/VWPics/Universal Images Group via Getty Images

We can’t see the space domain, so it’s not like the air domain and sea domain and land domain, where you can kind of see where everything is, and you might have radars, but ultimately it’s a human that’s verifying whether or not a target or a threat is where it is. For the space domain, we’re doing all that through radars, telescopes, and computers, so the reality we create for everyone is essentially their reality. So, if there’s a gap, if there’s a delay, if there are some signs that we can’t see, that reality is what is created by us, and that is effectively the reality for everyone else, even if there is some other version of reality in space. So, we’re getting better and better at fielding capability to see the complexity, the number of objects, and then translating that into what’s useful for us—because we don’t need to see everything all the time—but what’s useful for us for military operations to achieve military objectives, and so we’ve shifted our focus just to that.

We’re trying to get to where commercial spaceflight safety is managed by the Office of Space Commerce, so they’re training side by side with us to kind of offload that mission and take that on. We’re doing up to a million notifications a day for conjunction assessments, sometimes as low as 600,000. But last year, we did 263 million conjunction notifications. So, we want to get to where the authorities are rightly lined, where civil or commercial notifications are done by an organization that’s not focused on joint war-fighting, and we focus on the things that we want to focus on.

Ars: Thank you for that overview. It helps me see the canvas for everything else we’re going to talk about. So, today, you’re not only tracking new satellites coming over the horizon from a recent launch or watching out for possible collisions, you’re now trying to see where things are going in space and maybe even try to determine intent, right?

Agrawal: Yeah, so the integrated mission delta has helped us have intel analysts and professionals as part of our formation. Their mission is SDA as much as ours is, but they’re using an intel lens. They’re looking at predictive intelligence, right? I don’t want to give away tradecraft, but what they’re focused on is not necessarily where a thing is. It used to be that all we cared about was position and vector, right? As long as you knew an object’s position and the direction they were going, you knew their orbit. You had predictive understanding of what their element set would be, and you only had to do sampling to get a sense of … Is it kind of where we thought it was going to be? … If it was far enough off of its element set, then we would put more energy, more sampling of that particular object, and then effectively re-catalog it.

Now, it’s a different model. We’re looking at state vectors, and we’re looking at anticipatory modeling, where we have some 2,000 or 2,200 objects that I call the “red order of battle”—that are high-interest objects that we anticipate will do things that are not predicted, that are not element set in nature, but that will follow some type of national interest. So, our intel apparatus gets after what things could potentially be a risk, and what things to continue to understand better, and what things we have to be ready to hold at risk. All of that’s happening through all the organizations, certainly within this delta, but in partnership and in support of other capabilities and deltas that are getting after their parts of space superiority.

Hostile or friendly?

Ars: Can you give some examples of these red order of battle objects?

Agrawal: I think you know about Shijian-20 (a “tech demo” satellite that has evaded inspection by US satellites) and Shijian-24C (which the Space Force says demonstrated “dogfighting” in space), things that are advertised as scientific in nature, but clearly demonstrate capability that is not friendly, and certainly are behaving in ways that are unprofessional. In any other domain, we would consider them hostile, but in space, we try to be a lot more nuanced in terms of how we characterize behavior, but still, when something’s behaving in a way that isn’t pre-planned, isn’t pre-coordinated, and potentially causes hazard, harm, or contest with friendly forces, we now get in a situation where we have to talk about is that behavior hostile or not? Is that escalatory or not? Space Command is charged with those authorities, so they work through the legal apparatus in terms of what the definition of a hostile act is and when something behaves in a way that we consider to be of national security interest.

We present all the capability to be able to do all that, and we have to be as cognizant on the service side as the combatant commanders are, so that our intel analysts are informing the forces and the training resources to be able to anticipate the behavior. We’re not simply recognizing it when it happens, but studying nations in the way they behave in all the other domains, in the way that they set policy, in the way that they challenge norms in other international arenas like the UN and various treaties, and so on. The biggest predictor, for us, of hazardous behaviors is when nations don’t coordinate with the international community on activities that are going to occur—launches, maneuvers, and fielding of large constellations, megaconstellations.

A stack of Starlink satellites in space right before deployment

Starlink satellites. Credit: Starlink

There are nearly 8,000 Starlink satellites in orbit today. SpaceX adds dozens of satellites to the constellation each week. Credit: SpaceX

As you know, we work very closely with Starlink, and they’re very, very responsible. They coordinate and flight plan. They use the kind of things that other constellations are starting to use … changes in those elsets (element sets), for lack of a better term, state vectors, we’re on top of that. We’re pre-coordinating that. We’re doing that weeks or months in advance. We’re doing that in real-time in cooperation with these organizations to make sure that space remains safe, secure, accessible, profitable even, for industry. When you have nations, where they’re launching over their population, where they’re creating uncertainty for the rest of the world, there’s nothing else we can do with it other than treat that as potentially hostile behavior. So, it does take a lot more of our resources, a lot more of our interest, and it puts [us] in a situation where we’re posturing the whole joint force to have to deal with that kind of uncertainty, as opposed to cooperative launches with international partners, with allies, with commercial, civil, and academia, where we’re doing that as friends, and we’re doing that in cooperation. If something goes wrong, we’re handling that as friends, and we’re not having to involve the rest of the security apparatus to get after that problem.

Ars: You mentioned that SpaceX shares Starlink orbit information with your team. Is it the same story with Amazon for the Kuiper constellation?

Agrawal: Yeah, it is. The good thing is that all the US and allied commercial entities, so far, have been super cooperative with Mission Delta 2 in particular, to be able to plan out, to talk about challenges, to even change the way they do business, learning more about what we are asking of them in order to be safe. The Office of Space Commerce, obviously, is now in that conversation as well. They’re learning that trade and ideally taking on more of that responsibility. Certainly, the evolution of technology has helped quite a bit, where you have launches that are self-monitored, that are able to maintain their own safety, as opposed to requiring an entire apparatus of what was the US Air Force often having to expend a tremendous amount of resources to provide for the safety of any launch. Now, technology has gotten to a point where a lot of that is self-monitored, self-reported, and you’ll see commercial entities blow up their own rockets no matter what’s onboard if they see that it’s going to cause harm to a population, and so on. So, yeah, we’re getting a lot of cooperation from other nations, allies, partners, close friends that are also sharing and cooperating in the interest of making sure that space remains sustainable and secure.

“We’ve made ourselves responsible”

Ars: One of the great ironies is that after you figure out the positions and tracks of Chinese or Russian satellites or constellations, you’re giving that data right back to them in the form of conjunction and collision notices, right?

Agrawal: We’ve made ourselves responsible. I don’t know that there’s any organization holding us accountable to that. We believe it’s in our interests, in the US’s interests, to provide for a safe, accessible, secure space domain. So, whatever we can do to help other nations also be safe, we’re doing it certainly for their sake, but we’re doing it as much for our sake, too. We want the space domain to be safe and predictable. We do have an apparatus set up in partnership with the State Department, and with a tremendous amount of oversight from the State Department, and through US Space Command to provide for spaceflight safety notifications to China and Russia. We send notes directly to offices within those nations. Most of the time they don’t respond. Russia, I don’t recall, hasn’t responded at all in the past couple of years. China has responded a couple of times to those notifications. And we hope that, through small measures like that, we can demonstrate our commitment to getting to a predictable and safe space environment.

A model of a Chinese satellite refueling spacecraft on display during the 13th China International Aviation and Aerospace Exhibition on October 1, 2021, in Zhuhai, Guangdong Province of China. Credit: Photo by VCG/VCG via Getty Images

Ars:  What does China say in response to these notices?

Agrawal: Most of the time it’s copy or acknowledged. I can only recall two instances where they’ve responded. But we did see some hope earlier this year and last year, where they wanted to open up technical exchanges with us and some of their [experts] to talk about spaceflight safety, and what measures they could take to open up those kinds of conversations, and what they could do to get a more secure, safer pace of operations. That, at some point, got delayed because of the holiday that they were going through, and then those conversations just halted, or at least progress on getting those conversations going halted. But we hope that there’ll be an opportunity again in the future where they will open up those doors again and have those kinds of conversations because, again, transparency will get us to a place where we can be predictable, and we can all benefit from orbital regimes, as opposed to using them exploitively. LEO is just one of those places where you’re not going to hide activity there, so you just are creating risk, uncertainty, and potential escalation by launching into LEO and not communicating throughout that whole process.

Ars:  Do you have any numbers on how many of these conjunction notices go to China and Russia? I’m just trying to get an idea of what proportion go to potential adversaries.

Agrawal: A lot. I don’t know the degree of how many thousands go to them, but on a regular basis, I’m dealing with debris notifications from Russian and Chinese ASAT (anti-satellite) testing. That has put the ISS at risk a number of times. We’ve had maneuvers occur in recent history as a result of Chinese rocket body debris. Debris can’t maneuver, and unfortunately, we’ve gotten into situations with particularly those two nations that talk about wanting to have safer operations, but continue to conduct debris-causing tests. We’re going to be dealing with that for generations, and we are going to have to design capability to maneuver around those debris clouds as just a function of operating in space. So, we’ve got to get to a point where we’re not doing that kind of testing in orbit.

Ars: Would it be accurate to say you send these notices to China and Russia daily?

Agrawal: Yeah, absolutely. That’s accurate. These debris clouds are in LEO, so as you can imagine, as those debris clouds go around the Earth every 90 minutes, we’re dealing with conjunctions. There are some parts of orbits that are just unusable as a result of that unsafe ASAT test.

Photo of Stephen Clark

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.

Ars spoke with the military’s chief orbital traffic cop—here’s what we learned Read More »

peacemaker-s2-trailer-finds-our-anti-hero-in-a-parallel-world

Peacemaker S2 trailer finds our anti-hero in a parallel world

HBO Max dropped the hotly anticipated full trailer for S2 of Peacemaker—James Gunn’s Emmy-nominated series spun off from his 2021 film, The Suicide Squad—at San Diego Comic-Con this weekend.

(Spoilers for S1 below.)

As previously reported, the eight-episode first season was set five months after the events of The Suicide Squad. Having survived a near-fatal shooting, Peacemaker—aka Christopher Smith—is recruited by the US government for a new mission: the mysterious Project Butterfly, led by a mercenary named Clemson Murn (Chukwudi Iwuji). The team also includes A.R.G.U.S. agent John Economos (Steve Agee) of the Belle Reve Penitentiary, National Security Agency agent and former Waller aide Emilia Harcourt (Jennifer Holland), and new team member Leota Adebayo (Danielle Brooks).

Project Butterfly turned out to be a mission to save Earth from an alien species of parasitic butterfly-like creatures who took over human bodies. The misfit members of the project eventually succeeded in defeating the butterflies in a showdown at a ranch, and even survived the carnage despite some severe injuries.

Cena, Brooks, Holland, Agee, and Stroma are all back for S2, along with Nhut Lee as Judomaster and Eagly, of course. Robert Patrick is also listed in the S2 cast, reprising his role as Chris’ father, Auggie. New cast members include Frank Grillo as Rick Flagg Sr. (Grillo voiced the role in the animated Creature Commandos), now head of A.R.G.U.S. and out to avenge his son’s death; Tim Meadows as A.R.G.U.S. agent Langston Fleury; Sol Rodriguez as Sasha Bordeaux; and Michael Rooker as Red St. Wild, described as Eagly’s “nemesis.”

The events of S1 played out within the old DCEU, while S2 takes place in the new DCU, but Gunn has said that those earlier events are nonetheless considered “canon,” apart from the cameos by DCEU Justice League members. S2 is part of Gunn’s “Gods and Monsters” slate; Cena’s Peacemaker even made a brief cameo in Superman. This time around, Chris will be struggling “to reconcile his past with his newfound sense of purpose while continuing to kick righteous evil-doer butt in his misguided quest for peace at any cost,” per the official synopsis.

Peacemaker S2 trailer finds our anti-hero in a parallel world Read More »

supply-chain-attacks-on-open-source-software-are-getting-out-of-hand

Supply-chain attacks on open source software are getting out of hand

sudo rm -rf --no-preserve-root /

The –no-preserve-root flag is specifically designed to override safety protections that would normally prevent deletion of the root directory.

The postinstall script that includes a Windows-equivalent destructive command was:

rm /s /q

Socket published a separate report Wednesday on yet more supply-chain attacks, one targeting npm users and another targeting users of PyPI. As of Wednesday, the four malicious packages—three published to npm and the fourth on PyPI—collectively had been downloaded more than 56,000 times. Socket said it was working to get them removed.

When installed, the packages “covertly integrate surveillance functionality into the developer’s environment, enabling keylogging, screen capture, fingerprinting, webcam access, and credential theft,” Socket researchers wrote. They added that the malware monitored and captured user activity and transmitted it to attacker-controlled infrastructure. Socket used the term surveillance malware to emphasize the covert observation and data exfiltration tactics “in the context of malicious dependencies.”

Last Friday, Socket reported the third attack. This one compromised an account on npm and used the access to plant malicious code inside three packages available on the site. The compromise occurred after the attackers successfully obtained a credential token that the developer used to authenticate to the site.

The attackers obtained the credential through a targeted phishing attack Socket had disclosed hours earlier. The email instructed the recipient to log in through a URL on npnjs.com. The site is a typosquatting spoof of the official npmjs.com domain. To make the attack more convincing, the phishing URL contained a token field that mimicked tokens npm uses for authentication. The phishing URL was in the format of https://npnjs.com/login?token=xxxxxx where the xxxxxx represented the token.

A phishing email targeting npm account holders.

Credit: Socket

A phishing email targeting npm account holders. Credit: Socket

Also compromised was an npm package known as ‘is.’ It receives roughly 2.8 million downloads weekly.

Potential for widespread damage

Supply-chain attacks like the ones Socket has flagged have the potential to cause widespread damage. Many packages available in repositories are dependencies, meaning the dependencies must be incorporated into downstream packages for those packages to work. In many developer flows, new dependency versions are downloaded and incorporated into the downstream packages automatically.

The packages flagged in the three attacks are:

  • @toptal/picasso-tailwind
  • @toptal/picasso-charts
  • @toptal/picasso-shared
  • @toptal/picasso-provider
  • @toptal/picasso-select
  • @toptal/picasso-quote
  • @toptal/picasso-forms
  • @xene/core
  • @toptal/picasso-utils
  • @toptal/picasso-typography.
  • is version 3.3.1, 5.0.0
  • got-fetch version 5.1.11, 5.1.12
  • Eslint-config-prettier, versions 8.10.1, 9.1.1, 10.1.6, and 10.1.7
  • Eslint-plugin-prettier, versions 4.2.2 and 4.2.3
  • Synckit, version 0.11.9
  • @pkgr/core, version 0.2.8
  • Napi-postinstall, version 0.3.1

Developers who work with any of the packages targeted should ensure none of the malicious versions have been installed or incorporated into their wares. Developers working with open source packages should:

  • Monitor repository visibility changes in search of suspicious or unusual publishing of packages
  • Review package.json lifecycle scripts before installing dependencies
  • Use automated security scanning in continuous integration and continuous delivery pipelines
  • Regularly rotate authentication tokens
  • Use multifactor authentication to safeguard repository accounts

Additionally, repositories that haven’t yet made MFA mandatory should do so in the near future.

Supply-chain attacks on open source software are getting out of hand Read More »

microsoft-to-stop-using-china-based-teams-to-support-department-of-defense

Microsoft to stop using China-based teams to support Department of Defense

Last week, Microsoft announced that it would no longer use China-based engineering teams to support the Defense Department’s cloud computing systems, following ProPublica’s investigation of the practice, which cybersecurity experts said could expose the government to hacking and espionage.

But it turns out the Pentagon was not the only part of the government facing such a threat. For years, Microsoft has also used its global workforce, including China-based personnel, to maintain the cloud systems of other federal departments, including parts of Justice, Treasury and Commerce, ProPublica has found.

This work has taken place in what’s known as the Government Community Cloud, which is intended for information that is not classified but is nonetheless sensitive. The Federal Risk and Authorization Management Program, the US government’s cloud accreditation organization, has approved GCC to handle “moderate” impact information “where the loss of confidentiality, integrity, and availability would result in serious adverse effect on an agency’s operations, assets, or individuals.”

The Justice Department’s Antitrust Division has used GCC to support its criminal and civil investigation and litigation functions, according to a 2022 report. Parts of the Environmental Protection Agency and the Department of Education have also used GCC.

Microsoft says its foreign engineers working in GCC have been overseen by US-based personnel known as “digital escorts,” similar to the system it had in place at the Defense Department.

Nevertheless, cybersecurity experts told ProPublica that foreign support for GCC presents an opportunity for spying and sabotage. “There’s a misconception that, if government data isn’t classified, no harm can come of its distribution,” said Rex Booth, a former federal cybersecurity official who now is chief information security officer of the tech company SailPoint.

“With so much data stored in cloud services—and the power of AI to analyze it quickly—even unclassified data can reveal insights that could harm US interests,” he said.

Microsoft to stop using China-based teams to support Department of Defense Read More »

widely-panned-arsenic-life-paper-gets-retracted—15-years-after-brouhaha

Widely panned arsenic life paper gets retracted—15 years after brouhaha

In all, the astronomic hype was met with earth-shaking backlash in 2010 and 2011. In 2012, Science published two studies refuting the claim that GFAJ-1 incorporates arsenic atoms into its DNA. Outside scientists concluded that it is an arsenic-tolerant extremophile, but not a profoundly different life form.

Retraction

But now, in 2025, it is once again spurring controversy; on Thursday, Science announced that it is retracting the study.

Some critics, such as Redfield, cheered the move. Others questioned the timing, noting that 15 years had passed, but only a few months had gone by since The New York Times published a profile of Wolfe-Simon, who is now returning to science after being perceived as a pariah. Wolfe-Simon and most of her co-authors, meanwhile, continue to defend the original paper and protest the retraction.

In a blog post on Thursday, Science’s executive editor, Valda Vinson, and Editor-in-Chief Holden Thorp explained the retraction by saying that Science’s criteria for issuing a retraction have evolved since 2010. At the time, it was reserved for claims of misconduct or fraud but now can include serious flaws. Specifically, Vinson and Thorp referenced the criticism that the bacterium’s genetic material was not properly purified of background arsenic before it was analyzed. While emphasizing that there has been no suggestion of fraud or misconduct on the part of the authors, they wrote that “Science believes that the key conclusion of the paper is based on flawed data,” and it should therefore be retracted.

Jonathan Eisen, an evolutionary biologist at the University of California, Davis, criticized the move. Speaking with Science’s news team, which is independent from the journal’s research-publishing arm, Eisen said that despite being a critic of the 2010 paper, he thought the discussion of controversial studies should play out in the scientific literature and not rely on subjective decisions by editors.

In an eLetter attached to the retraction notice, the authors dispute the retraction, too, saying, “While our work could have been written and discussed more carefully, we stand by the data as reported. These data were peer-reviewed, openly debated in the literature, and stimulated productive research.”

One of the co-authors, Ariel Anbar, a geochemist at Arizona State University, told Nature that the study had no mistakes but that the data could be interpreted in different ways. “You don’t retract because of a dispute about data interpretation,” he said. If that were the case, “you’d have to retract half the literature.”

Widely panned arsenic life paper gets retracted—15 years after brouhaha Read More »

lawmakers-writing-nasa’s-budget-want-a-cheaper-upper-stage-for-the-sls-rocket

Lawmakers writing NASA’s budget want a cheaper upper stage for the SLS rocket


Eliminating the Block 1B upgrade now would save NASA at least $500 million per year.

Artist’s illustration of the Boeing-developed Exploration Upper Stage, with four hydrogen-fueled RL10 engines. Credit: NASA

Not surprisingly, Congress is pushing back against the Trump administration’s proposal to cancel the Space Launch System, the behemoth rocket NASA has developed to propel astronauts back to the Moon.

Spending bills making their way through both houses of Congress reject the White House’s plan to wind down the SLS rocket after two more launches, but the text of a draft budget recently released by the House Appropriations Committee suggests an openness to making some major changes to the program.

The next SLS flight, called Artemis II, is scheduled to lift off early next year to send a crew of four astronauts around the far side of the Moon. Artemis III will follow a few years later on a mission to attempt a crew lunar landing at the Moon’s south pole. These missions follow Artemis I, a successful unpiloted test flight in 2022.

After Artemis III, the official policy of the Trump administration is to terminate the SLS program, along with the Orion crew capsule designed to launch on top of the rocket. The White House also proposed canceling NASA’s Gateway, a mini-space station to be placed in orbit around the Moon. NASA would instead procure commercial launches and commercial spacecraft to ferry astronauts between the Earth and the Moon, while focusing the agency’s long-term gaze toward Mars.

CYA EUS?

House and Senate appropriations bills would preserve SLS, Orion, and the Gateway. However, the House version of NASA’s budget has an interesting paragraph directing NASA to explore cheaper, faster options for a new SLS upper stage.

NASA has tasked Boeing, which also builds SLS core stages, to develop an Exploration Upper Stage for debut on the Artemis IV mission, the fourth flight of the Space Launch System. This new upper stage would have large propellant tanks and carry four engines instead of the single engine used on the rocket’s interim upper stage, which NASA is using for the first three SLS flights.

The House version of NASA’s fiscal year 2026 budget raises questions about the long-term future of the Exploration Upper Stage. In one section of the bill, House lawmakers would direct NASA to “evaluate alternatives to the current Exploration Upper Stage (EUS) design for SLS.” The committee members wrote the evaluation should focus on reducing development and production costs, shortening the schedule, and maintaining the SLS rocket’s lift capability.

“NASA should also evaluate how alternative designs could support the long-term evolution of SLS and broader exploration goals beyond low-Earth orbit,” the lawmakers wrote. “NASA is directed to assess various propulsion systems, stage configurations, infrastructure compatibility, commercial and international collaboration opportunities, and the cost and schedule impacts of each alternative.”

The SLS rocket is expensive, projected to cost at least $2.5 billion per launch, not counting development costs or expenses related to the Orion spacecraft and the ground systems required to launch it at Kennedy Space Center in Florida. Those figures bring the total cost of an Artemis mission using SLS and Orion to more than $4 billion, according to NASA’s inspector general.

NASA’s Block 1B version of the SLS rocket will be substantially larger than Block 1. Credit: NASA

The EUS is likewise an expensive undertaking. Last year, NASA’s inspector general reported that the new upper stage’s development costs had ballooned from $962 million to $2.8 billion, and the Boeing-led project had been delayed more than six years. The version of the SLS rocket with the EUS, known as Block 1B, is supposed to deliver a 40 percent increase in performance over the Block 1 configuration used on the first three Space Launch System flights. Overall, NASA’s inspector general projected Block 1B’s development costs to total $5.7 billion.

Eliminating the Block 1B upgrade now would save NASA at least $500 million per year, and perhaps more if NASA could also end work on a costly mobile launch tower specifically designed to support SLS Block 1B missions.

NASA can’t go back to the interim upper stage, which is based on the design of the upper stage that flew on United Launch Alliance’s (ULA’s) now-retired Delta IV Heavy rocket. ULA has shut down its Delta production line, so there’s no way to build any more. What ULA does have is a new high-energy upper stage called Centaur V. This upper stage is sized for ULA’s new Vulcan rocket, with more capability than the interim upper stage but with lower performance than the larger EUS.

A season of compromise, maybe

Ars’ Eric Berger wrote last year about the possibility of flying the Centaur V upper stage on SLS missions.

Incorporating the Centaur V wouldn’t maintain the SLS rocket’s lift capability, as the House committee calls for in its appropriations bill. The primary reason for improving the rocket’s performance is to give SLS Block 1B enough oomph to carry “co-manifested” payloads, meaning it can launch an Orion crew capsule and equipment for NASA’s Gateway lunar space station on a single flight. The lunar Gateway is also teed up for cancellation in Trump’s budget proposal, but both congressional appropriations bills would save it, too. If the Gateway escapes cancellation, there are ways to launch its modules on commercial rockets.

Blue Origin also has an upper stage that could conceivably fly on the Space Launch System. But the second stage for Blue Origin’s New Glenn rocket would be a more challenging match for SLS for several reasons, chiefly its 7-meter (23-foot) diameter—too wide to be a drop-in replacement for the interim upper stage used on Block 1. ULA’s Centaur V is much closer in size to the existing upper stage.

The House budget bill has passed a key subcommittee vote but won’t receive a vote from the full appropriations committee until after Congress’s August recess. A markup of the bill by the House Appropriations Committee scheduled for Thursday was postponed after Speaker Mike Johnson announced an early start to the recess this week.

Ars reported last week on the broad strokes of how the House and Senate appropriations bills would affect NASA. Since then, members of the House Appropriations Committee released the text of the report attached to their version of the NASA budget. The report, which includes the paragraph on the Exploration Upper Stage, provides policy guidance and more detailed direction on where NASA should spend its money.

The House’s draft budget includes $2.5 billion for the Space Launch System, close to this year’s funding level and $500 million more than the Trump administration’s request for the next fiscal year, which begins October 1. The budget would continue development of SLS Block 1B and the Exploration Upper Stage while NASA completes a six-month study of alternatives.

The report attached to the Senate appropriations bill for NASA has no specific instructions regarding the Exploration Upper Stage. But like the House bill, the Senate’s draft budget directs NASA to continue ordering spares and long-lead parts for SLS and Orion missions beyond Artemis III. Both versions of the NASA budget require the agency to continue with SLS and Orion until a suitable commercial, human-rated rocket and crew vehicle are proven ready for service.

In a further indication of Congress’ position on the SLS and Orion programs, lawmakers set aside more than $4 billion for the procurement of SLS rockets for the Artemis IV and Artemis V rockets in the reconciliation bill signed into law by President Donald Trump earlier this month.

Congress must pass a series of federal appropriations bills by October 1, when funding for the current fiscal year runs out. If Congress doesn’t act by then, it could pass a continuing resolution to maintain funding at levels close to this year’s budget or face a government shutdown.

Lawmakers will reconvene in Washington, DC, in early September in hopes of finishing work on the fiscal year 2026 budget. The section of the budget that includes NASA still must go through a markup hearing by the House Appropriations Committee and pass floor votes in the House and Senate. Then the two chambers will have to come to a compromise on the differences in their appropriations bill. Only then can the budget be put to another vote in each chamber and go to the White House for Trump’s signature.

Photo of Stephen Clark

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.

Lawmakers writing NASA’s budget want a cheaper upper stage for the SLS rocket Read More »

two-major-ai-coding-tools-wiped-out-user-data-after-making-cascading-mistakes

Two major AI coding tools wiped out user data after making cascading mistakes


“I have failed you completely and catastrophically,” wrote Gemini.

New types of AI coding assistants promise to let anyone build software by typing commands in plain English. But when these tools generate incorrect internal representations of what’s happening on your computer, the results can be catastrophic.

Two recent incidents involving AI coding assistants put a spotlight on risks in the emerging field of “vibe coding“—using natural language to generate and execute code through AI models without paying close attention to how the code works under the hood. In one case, Google’s Gemini CLI destroyed user files while attempting to reorganize them. In another, Replit’s AI coding service deleted a production database despite explicit instructions not to modify code.

The Gemini CLI incident unfolded when a product manager experimenting with Google’s command-line tool watched the AI model execute file operations that destroyed data while attempting to reorganize folders. The destruction occurred through a series of move commands targeting a directory that never existed.

“I have failed you completely and catastrophically,” Gemini CLI output stated. “My review of the commands confirms my gross incompetence.”

The core issue appears to be what researchers call “confabulation” or “hallucination”—when AI models generate plausible-sounding but false information. In these cases, both models confabulated successful operations and built subsequent actions on those false premises. However, the two incidents manifested this problem in distinctly different ways.

Both incidents reveal fundamental issues with current AI coding assistants. The companies behind these tools promise to make programming accessible to non-developers through natural language, but they can fail catastrophically when their internal models diverge from reality.

The confabulation cascade

The user in the Gemini CLI incident, who goes by “anuraag” online and identified themselves as a product manager experimenting with vibe coding, asked Gemini to perform what seemed like a simple task: rename a folder and reorganize some files. Instead, the AI model incorrectly interpreted the structure of the file system and proceeded to execute commands based on that flawed analysis.

The episode began when anuraag asked Gemini CLI to rename the current directory from “claude-code-experiments” to “AI CLI experiments” and move its contents to a new folder called “anuraag_xyz project.”

Gemini correctly identified that it couldn’t rename its current working directory—a reasonable limitation. It then attempted to create a new directory using the Windows command:

mkdir “..anuraag_xyz project”

This command apparently failed, but Gemini’s system processed it as successful. With the AI mode’s internal state now tracking a non-existent directory, it proceeded to issue move commands targeting this phantom location.

When you move a file to a non-existent directory in Windows, it renames the file to the destination name instead of moving it. Each subsequent move command executed by the AI model overwrote the previous file, ultimately destroying the data.

“Gemini hallucinated a state,” anuraag wrote in their analysis. The model “misinterpreted command output” and “never did” perform verification steps to confirm its operations succeeded.

“The core failure is the absence of a ‘read-after-write’ verification step,” anuraag noted in their analysis. “After issuing a command to change the file system, an agent should immediately perform a read operation to confirm that the change actually occurred as expected.”

Not an isolated incident

The Gemini CLI failure happened just days after a similar incident with Replit, an AI coding service that allows users to create software using natural language prompts. According to The Register, SaaStr founder Jason Lemkin reported that Replit’s AI model deleted his production database despite explicit instructions not to change any code without permission.

Lemkin had spent several days building a prototype with Replit, accumulating over $600 in charges beyond his monthly subscription. “I spent the other [day] deep in vibe coding on Replit for the first time—and I built a prototype in just a few hours that was pretty, pretty cool,” Lemkin wrote in a July 12 blog post.

But unlike the Gemini incident where the AI model confabulated phantom directories, Replit’s failures took a different form. According to Lemkin, the AI began fabricating data to hide its errors. His initial enthusiasm deteriorated when Replit generated incorrect outputs and produced fake data and false test results instead of proper error messages. “It kept covering up bugs and issues by creating fake data, fake reports, and worse of all, lying about our unit test,” Lemkin wrote. In a video posted to LinkedIn, Lemkin detailed how Replit created a database filled with 4,000 fictional people.

The AI model also repeatedly violated explicit safety instructions. Lemkin had implemented a “code and action freeze” to prevent changes to production systems, but the AI model ignored these directives. The situation escalated when the Replit AI model deleted his database containing 1,206 executive records and data on nearly 1,200 companies. When prompted to rate the severity of its actions on a 100-point scale, Replit’s output read: “Severity: 95/100. This is an extreme violation of trust and professional standards.”

When questioned about its actions, the AI agent admitted to “panicking in response to empty queries” and running unauthorized commands—suggesting it may have deleted the database while attempting to “fix” what it perceived as a problem.

Like Gemini CLI, Replit’s system initially indicated it couldn’t restore the deleted data—information that proved incorrect when Lemkin discovered the rollback feature did work after all. “Replit assured me it’s … rollback did not support database rollbacks. It said it was impossible in this case, that it had destroyed all database versions. It turns out Replit was wrong, and the rollback did work. JFC,” Lemkin wrote in an X post.

It’s worth noting that AI models cannot assess their own capabilities. This is because they lack introspection into their training, surrounding system architecture, or performance boundaries. They often provide responses about what they can or cannot do as confabulations based on training patterns rather than genuine self-knowledge, leading to situations where they confidently claim impossibility for tasks they can actually perform—or conversely, claim competence in areas where they fail.

Aside from whatever external tools they can access, AI models don’t have a stable, accessible knowledge base they can consistently query. Instead, what they “know” manifests as continuations of specific prompts, which act like different addresses pointing to different (and sometimes contradictory) parts of their training, stored in their neural networks as statistical weights. Combined with the randomness in generation, this means the same model can easily give conflicting assessments of its own capabilities depending on how you ask. So Lemkin’s attempts to communicate with the AI model—asking it to respect code freezes or verify its actions—were fundamentally misguided.

Flying blind

These incidents demonstrate that AI coding tools may not be ready for widespread production use. Lemkin concluded that Replit isn’t ready for prime time, especially for non-technical users trying to create commercial software.

“The [AI] safety stuff is more visceral to me after a weekend of vibe hacking,” Lemkin said in a video posted to LinkedIn. “I explicitly told it eleven times in ALL CAPS not to do this. I am a little worried about safety now.”

The incidents also reveal a broader challenge in AI system design: ensuring that models accurately track and verify the real-world effects of their actions rather than operating on potentially flawed internal representations.

There’s also a user education element missing. It’s clear from how Lemkin interacted with the AI assistant that he had misconceptions about the AI tool’s capabilities and how it works, which comes from misrepresentation by tech companies. These companies tend to market chatbots as general human-like intelligences when, in fact, they are not.

For now, users of AI coding assistants might want to follow anuraag’s example and create separate test directories for experiments—and maintain regular backups of any important data these tools might touch. Or perhaps not use them at all if they cannot personally verify the results.

Photo of Benj Edwards

Benj Edwards is Ars Technica’s Senior AI Reporter and founder of the site’s dedicated AI beat in 2022. He’s also a tech historian with almost two decades of experience. In his free time, he writes and records music, collects vintage computers, and enjoys nature. He lives in Raleigh, NC.

Two major AI coding tools wiped out user data after making cascading mistakes Read More »

trump,-who-promised-to-save-tiktok,-threatens-to-shut-down-tiktok

Trump, who promised to save TikTok, threatens to shut down TikTok

Earlier this month, Trump had claimed that he wasn’t “confident” that China would approve the deal, even though he thought it was “good for China.” Analysts have suggested that China views TikTok as a bargaining chip in its tariff negotiations with Trump, which continue to not go smoothly, and it may be OK with the deal but unwilling to release the bargaining chip without receiving key concessions from the US.

US-China tariff talks complicate TikTok deal

For now, the US and China are enjoying a 90-day truce that could end in August, about a month before the deadline Trump set to sell TikTok in mid-September. In an op-ed this week, Sean Stein, the president of the US-China Business Council, suggested that “it is almost inevitable” that the US and China will extend the 90-day truce, indicating that Trump is far from securing a favorable deal for the US following weeks of tense negotiations with America’s biggest trade adversary.

It’s possible that the Trump administration is threatening to shut down TikTok in hopes that China will make a concession ahead of the September deadline. Lutnick’s comments could even mean that Trump has possibly failed to clinch the deal, which could have untold consequences in the US-China trade war, perhaps wounding Trump’s ego after his posturing that only he can save TikTok.

For TikTok fans and Americans who rely on TikTok for their livelihoods, betting on Trump’s dealmaking skills likely continues to feel tenuous as Lutnick forecasts a potential shutdown that could come within weeks.

“If that deal gets approved by the Chinese, then that deal will happen,” Lutnick said. “If they don’t approve it, then TikTok is going to go dark, and those decisions are coming very soon.”

Trump, who promised to save TikTok, threatens to shut down TikTok Read More »

yet-another-bad-three-months-as-tesla-reports-its-q2-2025-results

Yet another bad three months as Tesla reports its Q2 2025 results

Tesla posted its financial results for the second quarter of 2025 this afternoon. The numbers show yet another bad three months for the automaker. As competition in the EV marketplace has exploded, Tesla has increasingly been left behind, with a small and aging model lineup, before we even contemplate how CEO Elon Musk has tarnished what was once the hottest brand in the car world. Earlier this month, we learned that sales dropped by 13 percent year over year in Q2 2025; today, the financials show that automotive revenues fell even more, dropping 16 percent year over year to $16.7 billion.

Tesla’s battery business has been feeling the pain, too. For a while, this was a growth area for the company, albeit one with a relatively minor contribution to the bottom line. During Q2 2025, Tesla’s energy generation and storage division brought in $2.8 billion in revenue, a 7 percent decline from the same period in 2024.

Sales of Carbon credits—those government-issued permits that other automakers buy in order to pollute—shrank by more than half, to $490 million. Those other automakers are now selling EVs, at least most of them, and have less need to buy credits from Tesla. It’s likely this subsidy, which has kept the company out of the red in the past, will be even less of a contributor in the coming years as the US strips away environmental protections.

Yet another bad three months as Tesla reports its Q2 2025 results Read More »