Back in 2023 we noted how a company named Telly proclaimed it had come up with a new idea for a TV: a free TV, with a second small TV below it, that shows users ads pretty much all of the time. While the bottom TV could also be used for useful things (like weather or a stock tracker), the fact it was constantly bombarding you with ads was supposed to offset any need for a retail price.
But apparently there’s been trouble in innovation paradise.
Shortly after launch, Telly proclaimed that it expected to ship more than half a million of the ad-laden sets. Within a few months it had announced it had already received 250,000 pre-orders. But a recent report by Lowpass indicates that only 35,000 of the sets had made it to peoples’ homes.
What was the problem? Ars Technica, Lowpass and The Verge note that the problems began with a substandard shipping process that resulted in a lot of TVs showing up broken to folks who pre-ordered. Reddit is also full of complaints about general quality control issues, like color issues, ads being played too loudly, odd connectivity issues, remote controls randomly unpairing, and more.
Still, there’s evidence that the idea might still have legs, as the premise itself appears profitable:
“The investor update reportedly said Telly made $22 million in annualized revenue in Q3 2025. This could equate to about $52 in advertising revenue per Telly in use per month ($22 million divided by 35,000 TVs divided by 12 months in a year is $52.38).
That’s notably more than what other TV companies report, as Lowpass pointed out. As a comparison to other budget TV brands that rely heavily on ads and user tracking, Roku reported an average revenue per user (ARPU) of $41.49 for 2024. Vizio, meanwhile, reported an ARPU of $37.17 in 2024.”
The TV industry had already realized that they can make more money tracking your viewing and shopping behavior (and selling that information to dodgy data brokers) long term than they do on the retail value of the set. This just appears to be an extension of that concept, and if companies like Telly can get out of their own way on quality control, it’s likely you’ll see more of it.
In one sense that’s great if you can’t afford the newest and greatest TV set. It’s less great given that the United States is too corrupt to pass functional consumer privacy protections or keep its regulators staffed and functional, meaning there are increasingly fewer mechanisms preventing companies like this from exploiting all the microphone, input, and other data collected from users on a day-to-day basis.
I personally want the opposite experience; I’m willing to pay extra for a dumb television that’s little more than a display panel and some HDMI inputs. A device that has no real “smart” internals or bloated, badly designed GUI made by companies more interested in selling ads than quality control. Some business class TVs can sometimes fit the bill, but by and large it’s a segment the industry clearly isn’t interested in, because there’s much, much more money to be made spying on and monetizing your every decision.
In the grand scheme of things — the wanton cruelty, the routine violations of rights, the actual fucking murders — this may only seem like a blip on the mass deportation continuum. But this report from Dell Cameron for Wired is still important. It not only explains why federal officers are approaching people with cellphones drawn nearly as often as they’re approaching them with guns drawn, but also shows the administration is yet again pretending it’s a law unto itself.
On Wednesday, the Department of Homeland Security published new details about Mobile Fortify, the face recognition app that federal immigration agents use to identify people in the field, undocumented immigrants and US citizens alike. The details, including the company behind the app, were published as part of DHS’s 2025 AI Use Case Inventory, which federal agencies are required to release periodically.
The inventory includes two entries for Mobile Fortify—one for Customs and Border Protection (CBP), another for Immigration and Customs Enforcement (ICE)—and says the app is in the “deployment” stage for both. CBP says that Mobile Fortify became “operational” at the beginning of May last year, while ICE got access to it on May 20, 2025. That date is about a month before 404 Media first reported on the app’s existence.
A lot was going on last May, in terms of anti-migrant efforts and the casual refusal to recognize long-standing constitutional rights. That was the same month immigration officers were told they could enter people’s homes while only carrying self-issued “administrative warrants,” which definitely aren’t the same thing as the judicial warrants the government actually needs to enter areas provided the utmost in Fourth Amendment protection.
The app federal officers are using is made by NEC, a tech company that’s been around since long before ICE and CBP become the mobile atrocities they are. Prior to this revelation, NEC had only been associated with developing biometric software with an eye on crafting something that could be swiftly deployed and just as quickly scaled to meet the government’s needs. This particular app was never made public prior to this.
ICE claims it’s not a direct customer. It’s only a beneficiary of the CBP’s existing contract with NEC. That’s a meaningless distinction when multiple federal agencies have been co-opted into the administration’s bigoted push to rid the nation of brown people.
As is always the case (and this precedes Trump 2.0), CBP and ICE are rolling out tech far ahead of the privacy impact paperwork that’s supposed to filed before anything goes live.
While CBP says there are “sufficient monitoring protocols” in place for the app, ICE says that the development of monitoring protocols is in progress, and that it will identify potential impacts during an AI impact assessment. According to guidance from the Office of Management and Budget, which was issued before the inventory says the app was deployed for either CBP or ICE, agencies are supposed to complete an AI impact assessment before deploying any high-impact use case. Both CBP and ICE say the app is “high-impact” and “deployed.”
This is standard operating procedure for the federal government. The FBI and DEA were deploying surveillance tech well ahead of Privacy Impact Assessments (PIAs) as far back as [oh wow] 2014, while the nation was still being run by someone who generally appeared to be a competent statesman. That nothing has changed since makes it clear this problem is endemic.
But things are a bit worse now that Trump is running an administration stocked with fully-cooked MAGA acolytes. In the past, our rights might have received a bit of lip service and the occasional congressional hearing about the lack of required Privacy Impact Assessments.
None of that will be happening now. No one in the DHS is even going to bother to apply pressure to those charged with crafting these assessments. And no one will threaten (much less terminate) the tech deployment until these assessments have been completed. I would fully expect this second Trump term to come and go without the delivery of legally-required paperwork, especially since oversight of these agencies will be completely nonexistent as long as the GOP holds a congressional majority.
We lose. The freshly stocked swamp wins. And while it’s normal to expect the federal government to bristle at the suggestion of oversight, it’s entirely abnormal to allow an administration that embraces white Christian nationalism to act as though the only holy text any Trump appointee subscribes to was handed down by Aleister Crowley: Do what thou wilt. That is the whole of the law.
The Baton Rouge Police Department announced recently that it will begin using a drone designed by military equipment manufacturer Lockheed Martin and Edge Autonomy, making it one of the first local police departments to use an unmanned aerial vehicle (UAV) with a history of primary use in foreign war zones. Baton Rouge is now one of the first local police departments in the United States to deploy an unmanned aerial vehicle (UAV) with such extensive surveillance capabilities — a dangerous escalation in the militarization of local law enforcement.
This is a troubling development in an already long history of local law enforcement acquiring and utilizing military-grade surveillance equipment. It should be a cautionary tale that prods communities across the country to be proactive in ensuring that drones can only be acquired and used in ways that are well-documented, transparent, and subject to public feedback.
Drones can access and view spaces that are otherwise off-limits to law enforcement, including backyards, decks, and other areas of personal property.
Footage captured by camera-enabled drones may be stored and shared in ways that go far beyond the initial flight.
Additional camera-based surveillance can be installed on the drone, including automated license plate readers and the retroactive application of biometric analysis, such as face recognition.
However, the use of a military-grade drone hypercharges these concerns. Stalker VXE30’s surveillance capabilities extend for dozens of miles, and it can fly faster and longer than standard police drones already in use.
“It can be miles away, but we can still have a camera looking at your face, so we can use it for surveillance operations,” BRPD Police Chief TJ Morse told reporters.
Additionally troubling is the capacity to add additional equipment to these drones: so-called “payloads” that could include other types of surveillance equipment and even weapons.
The Baton Rouge community must put policies in place that restrict and provide oversight of any possible uses of this drone, as well as any potential additions law enforcement might make.
EFF has filed a public records request to learn more about the conditions of this acquisition and gaps in oversight policies. We’ve been tracking the expansion of police drone surveillance for years, and this acquisition represents a dangerous new frontier. We’ll continue investigating and supporting communities fighting back against the militarization of local police and mass surveillance. To learn more about the surveillance technologies being used in your city, please check out the Atlas of Surveillance.
It is no secret that large language models (LLMs) are being used routinely to modify and even write scientific papers. That’s not necessarily a bad thing: LLMs can help produce clearer texts with stronger logic, not least when researchers are writing in a language that is not their mother tongue. More generally, a recent analysis in Nature magazine, reported by Science magazine, found that scientists embracing AI — of any kind — “consistently make the biggest professional strides”:
AI adopters have published three times more papers, received five times more citations, and reach leadership roles faster than their AI-free peers.
But there is also a downside:
Not only is AI-driven work prone to circling the same crowded problems, but it also leads to a less interconnected scientific literature, with fewer studies engaging with and building on one another.
Another issue with LLMs, that of “hallucinated citations,” or “HalluCitations,” is well known. More seriously, entire fake publications can be generated using AI, and sold by so-called “paper mills” to unscrupulous scientists who wish to bolster their list of publications to help their career. In the field of biomedical research alone, a recent study estimated that over 100,000 fake papers were published in 2023. Not all of those were generated using AI, but progress in LLMs has made the process of creating fake articles much simpler.
Fake publications generated using LLMs are often obvious because of their lack of sophistication and polish. But a new service from OpenAI, called Prism, is likely to eliminate such easy-to-spot signs, by adding AI support to every aspect of writing a scientific paper:
Prism is a free workspace for scientific writing and collaboration, with GPT‑5.2—our most advanced model for mathematical and scientific reasoning—integrated directly into the workflow.
It brings drafting, revision, collaboration, and preparation for publication into a single, cloud-based, LaTeX-native workspace. Rather than operating as a separate tool alongside the writing process, GPT‑5.2 works within the project itself—with access to the structure of the paper, equations, references, and surrounding context.
It includes a number of features that make creating complex — and fake — papers extremely easy:
Search for and incorporate relevant literature (for example, from arXiv) in the context of the current manuscript, and revise text in light of newly identified related work
Create, refactor, and reason over equations, citations, and figures, with AI that understands how those elements relate across the paper
Turn whiteboard equations or diagrams directly into LaTeX, saving hours of time manipulating graphics pixel-by-pixel
There is even voice-based editing, allowing simple changes to be made without the need to write anything. But scientists are already worried that the power of OpenAI’s Prism will make a deteriorating situation worse. As an article on Ars Technica explains:
[Prism] has drawn immediate skepticism from researchers who fear the tool will accelerate the already overwhelming flood of low-quality papers into scientific journals. The launch coincides with growing alarm among publishers about what many are calling “AI slop” in academic publishing.
One field that is already plagued by AI slop is AI itself. An FT article on the topic points to an interesting attempt by the International Conference on Learning Representations (ICLR), a major gathering of researchers in the world of machine learning, to tackle this problem with punitive measures against authors and reviewers who violate the ICLR’s policies on LLM-generated material. For example:
Papers that make extensive usage of LLMs and do not disclose this usage will be desk rejected [that is, without sending them out for external peer review]. Extensive and/or careless LLM usage often results in false claims, misrepresentations, or hallucinated content, including hallucinated references. As stated in our previous blog post: hallucinations of this kind would be considered a Code of Ethics violation on the part of the paper’s authors. We have been desk -rejecting, and will continue to desk -reject, any paper that includes such issues.
Similarly:
reviewers [of submitted papers] are responsible for the content they post. Therefore, if they use LLMs, they are responsible for any issues in their posted review. Very poor quality reviews that feature false claims, misrepresentations or hallucinated references are also a code of ethics violation as expressed in the previous blog post. As such, reviewers who posted such poor quality reviews will also face consequences, including the desk rejection of their [own] submitted papers.
It is clearly not possible to stop scientists from using AI tools to check and improve their papers, nor should this be necessary, provided authors flag up such usage, and no errors are introduced as a result. A policy of the kind adopted by the ICLR requiring transparency about the extent to which AI has been used seems a sensible approach in the face of increasingly sophisticated tools like OpenAI’s Prism.
We’ve covered how there’s a real push afoot to implement statewide “right to repair” laws that try to make it cheaper, easier, and environmentally friendlier for you to repair the technology you own. Unfortunately, while all fifty states have at least flirted with the idea, only Massachusetts, New York, Texas, Minnesota, Colorado, California, Oregon, and Washington have actually passed laws.
And among those states, not one has actually enforced them despite a wide array of ongoing corporate offenses (though to be fair to states there is kind of a lot going on).
Of states that are looking to pass additional laws, Maine appears to be the closest, despite a lot of automaker lobbyist shenanigans. LD 1228, otherwise known as “An Act to Clarify Certain Terms in and to Make Other Changes to the Automotive Right to Repair Laws,” aims to make it easier and more affordable for Maine residents to repair what they own.
The reforms were approved by Maine voters as a ballot initiative in 2023, again displaying how these reforms see broad, bipartisan public support.
That gave the auto industry too much power over the reforms, so the bill in its current form was recently vetoed by Maine Governor Janet Mills. From her veto statement:
“This provision — which was notably not included in the Working Group’s unanimous recommendations — was included at the urging of automobile manufacturers. However, without timely access to vehicle data, independent auto shops are left at a significant competitive disadvantage, and consumers would have fewer choices for automotive service and repair. With this provision included, LD 1228 would undermine the existing law overwhelmingly approved by Maine voters and harm independent repair shops across the state.”
The House, being pressured by automaker lobbyists, over-rode Mills’ veto, but the Senate flipped and upheld the veto after automakers went overtime spreading scary (and false) stories about how right to repair reforms pose a dire new security and privacy risks (they don’t). In some states, automakers have even lied and claimed that such reforms are a boon to sexual predators.
It’s another example of how, while we’re supposed to function as a representative democracy, corruption ensures that passing positive and even hugely popular reforms that challenge entrenched corporate power is as difficult as possible. And even if Maine does get a useful bill passed, serious enforcement is still an open question given limited state resources and attention spans in the Trump era.
Let’s start with Flock, the company behind a number of automated license plate reader (ALPR) and other camera technologies. You might be surprised at how many Flock cameras there are in your community. Many large and small municipalities around the country have signed deals with Flock for license plate readers to track the movement of all cars in their city. Even though these deals are signed by local police departments, oftentimes ICE also gains access.
Because of their ubiquity, people are interested in finding out where and how many Flock cameras are in their community. One project that can help with this is the OUI-SPY, a small piece of open source hardware. The OUI-SPY runs on a cheap Arduino compatible chip called an ESP-32. There are multiple programs available for loading on the chip, such as “Flock You,” which allows people to detect Flock cameras and “Sky-Spy” to detect overhead drones. There’s also “BLE Detect,” which detects various Bluetooth signals including ones from Axon, Meta’s Ray-Bans that secretly record you, and more. It also has a mode commonly known as “fox hunting” to track down a specific device. Activists and researchers can use this tool to map out different technologies and quantify the spread of surveillance.
There’s also the open source Wigle app which is primarily designed for mapping out Wi-Fi, but also has the ability to make an audio alert when a specific Wi-Fi or Bluetooth identifier is detected. This means you can set it up to get a notification when it detects products from Flock, Axon, or other nasties in their vicinity.
One enterprising YouTuber, Benn Jordan, figured out a way to fool Flock cameras into not recording his license plate simply by painting some minor visual noise on his license plate. This is innocuous enough that any human will still be able to read his license plate, but it completely prevented Flock devices from recognizing his license plate as a license plate at the time. Some states have outlawed drivers obscuring their license plates, so taking such action is not recommended.
Jordan later went on to discover hundreds of misconfigured Flock cameras that were exposing their administrator interface without a password on the public internet. This would allow anyone with an internet connection to view a live surveillance feed, download 30 days of video, view logs, and more. The cameras pointed at parks, public trails, busy intersections, and even a playground. This was a massive breach of public trust and a huge mistake for a company that claims to be working for public safety.
Other hackers have taken on the task of open-source intelligence and community reporting. One interesting example is deflock.me and alpr.watch, which are crowdsourced maps of ALPR cameras. Much like the OUI-SPY project, this allows activists to map out and expose Flock surveillance cameras in their community.
Another interesting project documenting ICE and creating a trove of open-source intelligence is ICE List Wiki which contains info on companies that have contracts with ICE, incidents and encounters with ICE, and vehicles ICE uses.
People without programming knowledge can also get involved. In Chicago, people used whistles to warn their neighbors that ICE was present or in the area. Many people 3D-printed whistles along with instructional booklets to hand out to their communities, allowing a wider distribution of whistles and consequently earlier warnings for their neighbors.
There is also EFF’s own Rayhunter project for detecting cell-site simulators, about which we have written extensively. Rayhunter runs on a cheap mobile hotspot and doesn’t require deep technical knowledge to use.
It’s important to remember that we are not powerless. Even in the face of a domestic law enforcement presence with massive surveillance capabilities and military-esque technologies, there are still ways to engage in surveillance self-defense. We cannot give into nihilism and fear. We must continue to find small ways to protect ourselves and our communities, and when we can, fight back.
EFF is not affiliated with any of these projects (other than Rayhunter) and does not endorse them. We don’t make any statements about the legality of using any of these projects. Please consult with an attorney to determine what risks there may be.
In my previous posts about the use of generative AI tools in the video game industry, I have tried to drive home the point that a nuanced conversation is needed here. Predictably, there were many comments of the sort of stratified opinions that I was specifically attempting to avoid, but I always knew they’d be there. And that’s okay. Where there is novelty, there is disruption and discomfort. And, frankly, some of the dangers here aren’t unfounded.
But in the end, I remain of the opinion that generative AI will be a tool used by game developers generally in the future, if not the present. I also still firmly believe that the conversation we should be having is not whether AI should be used in games, but how it should be used.
And people like the CEO of Shift Up in South Korea sure aren’t helping when they insist on the need to use AI by trotting out the Chinese boogeyman.
Will gen AI be part of Stellar Blade 2‘s development? It doesn’t sound entirely outside the realm of possibility after recent comments from developer Shift Up’s CEO. The South Korean game studio is currently working on a sequel to the 2024 sci-fi action game and its boss thinks AI is the only way to compete with the massive development teams coming out of China.
“We devote around 150 people to a single game, but China puts in between 1,000 to 2,000,” Hyung-tae Kim, who also served as director on Stellar Blade, said during a recent conference briefing according to GameMeca (translated via Automaton). “We lack the capacity to compete, both in terms of quality and volume of content.”
Where do I even begin with this nonsense? First, it’s completely devoid of the nuance I was asking for in these kinds of discussions. This is essentially stating that developers can make up for China’s massive human assets it can throw at game development by using AI to make up the difference. 1 employee using AI, doing the math, can be the equivalent of 100 or so Chinese workers. That sounds like you’re looking to stave off hiring by using AI and you aren’t helping!
It also fails, somehow, to recognize that generative AI can be used in China as well. China isn’t exactly ignoring AI tools, you know, so this arms race makes no real sense.
Finally, it’s just kind of bullshit. Chinese studios have certainly produced some games, some that have been quite successful. But when we think about the major players in the video game industry, especially in terms of quality and revenue, China is but a fairly average player on the world scene. Tencent, NetEase, and MiHoYo all crack the top ten in revenue, but the rest of the longer list is filled with American, Japanese, and South Korean studios, among some other countries. They’re a player in the industry, to be sure. But they aren’t some dominant force that requires special tactics to compete with.
But despite all the above, Shift Up has been both successful and has committed to retaining and treating its staff well.
Was Kim actually worried about rising competition from China, or was he just flexing his geopolitical muscle as Stellar Blade‘s popularity catapults Shift Up into the big time? After all, that game sold millions of copies across console and PC without the help of AI, even as Tencent, Net Ease, and other major Chinese publishers flood the market with AAA free-to-play games.
For now at least, Shift Up employees are being well taken care of. Seoul Economic Daily recently reported that all 300 employees at the studio were given AirPods Max, Apple Watches, and a bonus $3,400 to celebrate the company’s profitable 2025. Why no video game consoles? It already gifted PS5 Pros and Switch 2s last year.
That sure doesn’t read like a studio in dire straits due to the scary Big Red Machine or whatever he’s trying to pitch. How about you keep making good games and all will be fine?
Then we can get back to the real, more nuanced conversation about just what place AI has in video game production.
Last week we noted how the Trump FCC, at the direct request of wireless phone giants, destroyed popular rules making it easier and cheaper to switch wireless carriers. The rules, applied via spectrum acquisition and merger conditions, required that Verizon unlock your phone within 60 days after purchase so you could easily switch to competitors.
Verizon, as we’ve long established, hates competition, and immediately got to work lobbying the Trump administration to destroy the rules. The pay-to-play Trump administration quickly agreed, and now Verizon has started telling wireless customers they have to wait a year before switching phones after purchasing one from Verizon:
“Verizon was previously required to unlock phones automatically after 60 days due to restrictions imposed on its spectrum licenses and merger conditions that helped Verizon obtain approval of its purchase of TracFone. But an update applied today to the TracFone unlocking policy said new phones will be locked for at least a year and that each customer will have to request an unlock instead of getting it automatically.”
Again, these conditions were broadly popular and served the public interest, ensuring that it was easier for consumers to switch between our ever-consolidating, anti-competitive wireless phone giants. Verizon lobbied the FCC by repeatedly lying, without evidence, that these conditions resulted in a wave of black market phone thefts. FCC boss Brendan Carr, ever the industry lackey, parroted the claims in his rulings.
To be clear this is, for now, only something Verizon is doing via its prepaid sub-brands that include Straight Talk, Tracfone, Net10 Wireless, Clearway, Total Wireless, Simple Mobile, SafeLink Wireless, and Walmart Family Mobile. These brands often attract lower income customers who can least afford to be trapped under an expensive provider like this.
You can, for now, still buy an unlocked phone from an independent retailer, bring it to Verizon’s main postpaid brands, and port it back out again if you’d like. But when Verizon sees limited Democrat and press backlash to this first push (guaranteed with so much else going on), it will steadily keep expanding its restrictions to include its primary brands and all unlocked phones.
I know this because I’ve covered this company for a quarter century and this company’s anti-competitive ambitions are as predictable as the tides.
Ideally, Verizon wants to return to what it considers the golden era of cellular phones: circa 2007 or so when carriers restricted how you could use your phone and restricted what apps you could install (remember all the shitty VCast Verizon apps they wouldn’t let you uninstall? Or the way they’d block phone GPS hardware from working on third-party apps?). Back then, they would also tether you to one carrier via expensive long-term contracts with costly early termination fees.
If we stay on this path of zero U.S. corporate oversight, it’s all coming back, sooner or later. From there, should U.S. governance remain under corrupt authoritarian dominance, it’s only a matter of time before Verizon tries to dictate what content you can see in collaboration with the kakistocracy, thanks to the Trump administration’s destruction of popular net neutrality protections.
This has always been Verizon’s ambition as a lumbering telecom giant that can’t innovate and hates competition and government oversight. Thanks to Trump’s assault on regulators, it’s increasingly difficult to hold companies like AT&T and Verizon accountable for literally anything (see the 5th Circuit’s decision to let AT&T off the hook for lying to, and spying on, its users).
And the Trump administration’s ongoing quest to rubber stamp every merger that comes across its desk means more consolidation, and ultimately higher prices for U.S. wireless consumers who already pay some of the highest prices for mobile data in the developed world.
Verizon and other broadly despised telecoms have struck a generational blow against oversight and consumer protection across Trump’s two terms, and they intend to take full advantage of a presidency they helped purchase. All while the president informs his loyal rubes he’s a champion of affordability.
This story was originally published by ProPublica.Republished under a CC BY-NC-ND 3.0license.There are additional (exceptional!) imagery in the original.
When SpaceX CEO Elon Musk chose a remote Texas outpost on the Gulf Coast to develop his company’s ambitious Starship, he put the 400-foot rocket on a collision course with the commercial airline industry.
Each time SpaceX did a test run of Starship and its booster, dubbed Super Heavy, the megarocket’s flight path would take it soaring over busy Caribbean airspace before it reached the relative safety of the open Atlantic Ocean. The company planned as many as five such launches a year as it perfected the craft, a version of which is supposed to one day land on the moon.
The FAA, which also oversees commercial space launches, predicted the impact to the national airspace would be “minor or minimal,” akin to a weather event, the agency’s 2022 approval shows. No airport would need to close and no airplane would be denied access for “an extended period of time.”
But the reality has been far different. Last year, three of Starship’s five launches exploded at unexpected points on their flight paths, twice raining flaming debris over congested commercial airways and disrupting flights. And while no aircraft collided with rocket parts, pilots were forced to scramble for safety.
A ProPublica investigation, based on agency documents, interviews with pilots and passengers, air traffic control recordings and photos and videos of the events, found that by authorizing SpaceX to test its experimental rocket over busy airspace, the FAA accepted the inherent risk that the rocket might put airplane passengers in danger.
And once the rocket failed spectacularly and that risk became real, neither the FAA nor Secretary of Transportation Sean Duffy sought to revoke or suspend Starship’s license to launch, a move that is permitted when “necessary to protect the public health and safety.” Instead, the FAA allowed SpaceX to test even more prototypes over the same airspace, adding stress to the already-taxed air traffic control system each time it launched.
The first two Starship explosions last year forced the FAA to make real-time calls on where to clear airspace and for how long. Such emergency closures camewith little or no warning, ProPublica found, forcing pilots to suddenly upend their flight plans and change course in heavily trafficked airspace to get out of the way of falling debris. In one case, a plane with 283 people aboard ran low on fuel, prompting its pilot to declare an emergency and cross a designated debris zone to reach an airport.
The world’s largest pilots union told the FAA in October that such events call into question whether “a suitable process” is in place to respond to unexpected rocket mishaps.
“There is high potential for debris striking an aircraft resulting in devastating loss of the aircraft, flight crew, and passengers,” wrote Steve Jangelis, a pilot and aviation safety chair.
The FAA said in response to questions that it “limits the number of aircraft exposed to the hazards, making the likelihood of a catastrophic event extremely improbable.”
Yet for the public and the press, gauging that danger has been difficult. In fact, nearly a year after last January’s explosion, it remains unclear just how close Starship’s wreckage came to airplanes. SpaceX estimated where debris fell after each incident and reported that information to the federal government. But the company didn’t respond to ProPublica’s requests for that data, and the federal agencies that have seen it, including the FAA, haven’t released it. The agency told us that it was unaware of any other publicly available data on Starship debris.
In public remarks, Musk downplayed the risk posed by Starship. To caption a video of flaming debris in January, he wrote, “Entertainment is guaranteed!” and, after the March explosion, he posted, “Rockets are hard.” The company has been more measured, saying it learns from mistakes, which “help us improve Starship’s reliability.”
For airplanes traveling at high speeds, there is little margin for error. Research shows as little as 300 grams of debris — or two-thirds of a pound — “could catastrophically destroy an aircraft,” said Aaron Boley, a professor at the University of British Columbia who has studied the danger space objects pose to airplanes. Photographs of Starship pieces that washed up on beaches show items much bigger than that, including large, intact tanks.
“It doesn’t actually take that much material to cause a major problem to an aircraft,” Boley said.
In response to growing alarm over the rocket’s repeated failures, the FAA has expanded prelaunch airspace closures and offered pilots more warning of potential trouble spots. The agency said it also required SpaceX to conduct investigations into the incidents and to “implement numerous corrective actions to enhance public safety.” An FAA spokesperson referred ProPublica’s questions about what those corrective actions were to SpaceX, which did not respond to multiple requests for comment.
Experts say the FAA’s shifting approach telegraphs a disquieting truth about air safety as private companies increasingly push to use the skies as their laboratories: Regulators are learning as they go.
During last year’s Starship launches, the FAA was under pressure to fulfill a dual mandate: to regulate and promote the commercial space industry while keeping the flying public safe, ProPublica found. In his October letter, Jangelis called the arrangement “a direct conflict of interest.”
In an interview, Kelvin Coleman, who was head of FAA’s commercial space office during the launches, said his office determined that the risk from the mishaps “was within the acceptable limits of our regulations.”
But, he said, “as more launches are starting to take place, I think we have to take a real hard look at the tools that we have in place and how do we better integrate space launch into the airspace.”
“We Need to Protect the Airspace”
On Jan. 16, 2025, as SpaceX prepared to launch Starship 7 from Boca Chica, Texas, the government had to address the possibility the giant rocket would break up unexpectedly.
Using debris modeling and simulations, the U.S. Space Force, the branch of the military that deals with the nation’s space interests, helped the FAA draw the contours of theoretical “debris response areas” — no-fly zones that could be activated if Starship exploded.
With those plans in place, Starship Flight 7 lifted off at 5:37 p.m. EST. About seven minutes later, it achieved a notable feat: Its reusable booster rocket separated, flipped and returned to Earth, where giant mechanical arms caught it as SpaceX employees cheered.
But about 90 seconds later, as Starship’s upper stage continued to climb, SpaceX lost contact with it. The craft caught fire and exploded, far above Earth’s surface.
A pilot on a flight from Miami to Santo Domingo, Dominican Republic, recorded video of space debris visible from the cockpit while flying at 37,000 feet. Provided to ProPublica
Air traffic control’s communications came alive with surprised pilots who saw the accident, some of whom took photos and shot videos of the flaming streaks in the sky:
Pilot: I just got a major streak going for at least 60 miles, all these different colors. Just curious but — it looked like it was coming towards us, but obviously because of the distance …. Just letting you know. Controller: Can you, can you give an estimate on how far away it is?
Another controller warned a different pilot of debris in the area:
Controller: Due to a space vehicle mishap — a rocket launch that basically exploded between our airspace and Miami — I’m going to give you holding instructions because there was debris in the area, so I’m going to keep you away from it.
Two FAA safety inspectors were in Boca Chica to watch the launch at SpaceX’s mission control, said Coleman, who, for Flight 7, was on his laptop in Washington, D.C., receiving updates.
As wreckage descended rapidly toward airplanes’ flight paths over the Caribbean, the FAA activated a no-fly zone based on the vehicle’s last known position and prelaunch calculations. Air traffic controllers warned pilots to avoid the area, which stretched hundreds of miles over a ribbon of ocean roughly from the Bahamas to just east of St. Martin, covering portions of populated islands, including all of Turks and Caicos. While the U.S. controls some airspace in the region, it relies on other countries to cooperate when it recommends a closure.
The FAA also cordoned off a triangular zone south of Key West.
When a pilot asked when planes would be able to proceed through the area, a controller replied:
Controller: The only information I got is that the rocket exploded so we need to protect the airspace, and Miami and Domingo stopped taking aircraft.
There were at least 11 planes in the closed airspace when Starship exploded, and flight tracking data shows they hurried to move out of the way, clearing the area within 15 minutes. Such maneuvers aren’t without risk. “If many aircraft need to suddenly change their routing plans,” Boley said, “then it could cause additional stress” on an already taxed air traffic control system, “which can lead to errors.”
That wasn’t the end of the disruption though. The FAA kept the debris response area, or DRA, active for another 71 minutes, leaving some flights in a holding pattern over the Caribbean. Several began running low on fuel and some informed air traffic controllers that they needed to land.
“We haven’t got enough fuel to wait,” said one pilot for Iberia airlines who was en route from Madrid with 283 people on board.
The controller warned him that if he proceeded across the closed airspace, it would be at his own risk:
Controller: If you’re going to pass through the DRA, you guys’re going to need to declare an emergency. That’s what my supervisor — if you’re going to land at San Juan, you need to declare an emergency for fuel reasons, that’s what my supervisor just told me. Pilot: In that case, we declare emergency. Mayday mayday mayday.
The plane landed safely in San Juan, Puerto Rico.
Iberia did not respond to requests for comment, but in statements to ProPublica, other airlines downplayed the launch fallout. Delta, for example, said the incident “had minimal impact to our operation and no aircraft damage.” The company’s “safety management system and our safety culture help us address potential issues to reinforce that air transportation remains the safest form of travel in the world,” a spokesperson said.
After the incident, some pilots registered concerns with the FAA, which was also considering a request from SpaceX to increase the number of annual Starship launches from five to 25.
“Last night’s Space X rocket explosion, which caused the diversion of several flights operating over the Gulf of Mexico, was pretty eye opening and scary,” wrote Steve Kriese in comments to the FAA, saying he was a captain for a major airline and often flew over the Gulf. “I do not support the increase of rocket launches by Space X, until a thorough review can be conducted on the disaster that occurred last night, and safety measures can be put in place that keeps the flying public safe.”
Kriese could not be reached for comment.
The Air Line Pilots Association urged the FAA to suspend Starship testing until the root cause of the failure could be investigated and corrected. A letter from the group, which represents more than 80,000 pilots flying for 43 airlines, said flight crews traveling in the Caribbean didn’t know where planes might be at risk from rocket debris until after the explosion.
“By that time, it’s much too late for crews who are flying in the vicinity of the rocket operation, to be able to make a decision for the safe outcome of the flight,” wrote Jangelis, the pilot and aviation safety chair for the group. The explosion, he said, “raises additional concerns about whether the FAA is providing adequate separation of space operations from airline flights.”
In response, the FAA said it would “review existing processes and determine whether additional measures can be taken to improve situational awareness for flight crews prior to launch.”
According to FAA documents, the explosion propelled Starship fragments across an area nearly the size of New Jersey. Debris landed on beaches and roadways in Turks and Caicos. It also damaged a car. No one was injured.
Three months later, the National Oceanic and Atmospheric Administration, which was evaluating potential impacts to marine life, sent the FAA a report with a map of where debris from an explosion could fall during future Starship failures. The estimate, which incorporated SpaceX’s own data from the Starship 7 incident, depicted an area more than three times the size of the airspace closed by the FAA.
In a statement, an FAA spokesperson said NOAA’s map was “intended to cover multiple potential operations,” while the FAA’s safety analysis is for a “single actual launch.” A NOAA spokesperson said that the map reflects “the general area where mishaps could occur” and is not directly comparable with the FAA’s no-fly zones.
Nevertheless Moriba Jah, a professor of aerospace engineering at the University of Texas, said the illustration suggested the no-fly zones the FAA activated may not fully capture how far and wide debris spreads after a rocket breakup. The current predictive science, he said, “carries significant uncertainty.”
At an industry conference a few weeks after the January explosion, Shana Diez, a SpaceX executive, acknowledged the FAA’s challenges in overseeing commercial launches.
“The biggest thing that we really would like to work with them on in the future is improving their real time awareness of where the launch vehicles are and where the launch vehicles’ debris could end up,” she said.
“We’re Too Close to the Debris”
On Feb. 26 of last year, with the investigation into Starship Flight 7 still open, the FAA cleared Flight 8 to proceed, saying it “determined SpaceX met all safety, environmental and other licensing requirements.”
The action was allowed under a practice that began during the first Trump administration, known as “expedited return-to-flight,” that permitted commercial space companies to launch again even before the investigation into a prior problematic flight was complete, as long as safety systems were working properly.
Coleman, who took a voluntary separation offer last year, said that before granting approval, the FAA confirmed that “safety critical systems,” such as the rocket’s ability to self-destruct if it went off course, worked as designed during Flight 7.
By March 6, SpaceX was ready to launch again. This time the FAA gave pilots a heads-up an hour and 40 minutes before liftoff.
“In the event of a debris-generating space launch vehicle mishap, there is the potential for debris falling within an area,” the advisory said, again listing coordinates for two zones in the Gulf and Caribbean.
The FAA said a prelaunch safety analysis, which includes planning for potential debris, “incorporates lessons learned from previous flights.” The zone described in the agency’s advisory for the Caribbean was wider and longer than the previous one, while the area over the Gulf was significantly expanded.
Flight 8 launched at 6:30 p.m. EST and its booster returned to the launchpad as planned. But a little more than eight minutes into the flight, some of Starship’s engines cut out. The craft went into a spin and about 90 seconds later SpaceX lost touch with it and it exploded.
The FAA activated the no-fly zones less than two minutes later, using the same coordinates it had released prelaunch.
Even with the advance warning, data shows at least five planes were in the debris zones at the time of the explosion, and they all cleared the airspace in a matter of minutes.
A pilot on one of those planes, Frontier Flight 081, told passengers they could see the rocket explosion out the right-side windows. Dane Siler and Mariah Davenport, who were heading home to the Midwest after vacationing in the Dominican Republic, lifted the window shade and saw debris blazing across the sky, with one spot brighter than the rest.
“It literally looked like the sun coming out,” Siler told ProPublica. “It was super bright.”
They and other passengers shot videos, marveling at what looked like fireworks, the couple said. The Starship fragments appeared to be higher than the plane, many miles off. But before long, the pilot announced “I’m sorry to report that we have to turn around because we’re too close to the debris,” Siler said.
Frontier did not respond to requests for comment.
The FAA lifted the restriction on planes flying through the debris zone about 30 minutes after Starship exploded, much sooner than it had in January. The agency said that the Space Force had “notified the FAA that all debris was down approximately 30 minutes after the Starship Flight 8 anomaly.”
But in response to ProPublica’s questions, the Space Force acknowledged that it did not track the debris in real time. Instead, it said “computational modeling,” along with other scientific measures, allowed the agency to “predict and mitigate risks effectively.” The FAA said “the aircraft were not at risk” during the aftermath of Flight 8.
Experts told ProPublica that the science underlying such modeling is far from settled, and the government’s ability to anticipate how debris will behave after an explosion like Starship’s is limited. “You’re not going to find anybody who’s going to be able to answer that question with any precision,” said John Crassidis, an aerospace engineering professor at the University of Buffalo. “At best, you have an educated guess. At worst, it’s just a potshot.”
Where pieces fall — and how long they take to land — depends on many factors, including atmospheric winds and the size, shape and type of material involved, experts said.
During the breakup of Flight 7, the FAA kept airspace closed for roughly 86 minutes. However, Diez, the SpaceX executive, told attendees at the industry conference that, in fact, it had taken “hours” for all the debris to reach the ground. The FAA, SpaceX and Diez did not respond to follow-up questions about her remarks.
It’s unclear how accurate the FAA’s debris projections were for the March explosion. The agency acknowledged that debris fell in the Bahamas, but it did not provide ProPublica the exact location, making it impossible to determine whether the wreckage landed where the FAA expected. While some of the country’s islands were within the boundaries of the designated debris zone, most were not. Calls and emails to Bahamas officials were not returned.
The FAA said no injuries or serious property damage occurred.
FAA Greenlights More Launches
By May, after months of Musk’s Department of Government Efficiency slashing spending and firing workers at federal agencies across Washington, the FAA granted SpaceX’s request to exponentially increase the number of Starship launches from Texas.
Starship is key to “delivering greater access to space and enabling cost-effective delivery of cargo and people to the Moon and Mars,” the FAA found. The agency said it will make sure parties involved “are taking steps to ensure the safe, efficient, and equitable use” of national airspace.
The U.S. is in a race to beat China to the lunar surface — a priority set by Trump’s first administration and continued under President Joe Biden. Supporters say the moon can be mined for resources like water and rare earth metals, and can offer a place to test new technologies. It could also serve as a stepping stone for more distant destinations, enabling Musk to achieve his longstanding goal of bringing humans to Mars.
Trump pledged last January that the U.S. will “pursue our Manifest Destiny into the stars, launching American astronauts to plant the Stars and Stripes on the planet Mars.”
But with experimental launches like Starship’s, Jangelis said, the FAA should be “as conservative as possible” when managing the airspace below them.
“We expect the FAA to make sure our aircraft and our passengers stay safe,” he said. “There has to be a balance between the for-profit space business and the for-profit airlines and commerce.”
A More Conservative Approach
In mid-May, United Kingdom officials sent a letter to their U.S. counterparts, asking that SpaceX and the FAA change Starship’s flight path or take other precautions because they were worried about the safety of their Caribbean territories.
The following day, the FAA announced in a news release that it had approved the next Starship launch, pending either the agency’s closure of the investigation into Flight 8 or granting of a “return to flight” determination.
A week later, with the investigation into Flight 8 still open, the agency said SpaceX had “satisfactorily addressed” the causes of the mishap. The FAA did not detail what those causes were at the time but said it would verify that the company implemented all necessary “corrective actions.”
This time the FAA was more aggressive on air safety.
The agency preventively closed an extensive swath of airspace extending 1,600 nautical miles from the launch site, across the Gulf of Mexico and through part of the Caribbean. The FAA said that 175 flights or more could be affected, and it advised Turks and Caicos’ Providenciales International Airport to close during the launch.
The agency said the move was driven in part by an “updated flight safety analysis” and SpaceX’s decision to reuse a previously launched Super Heavy booster — something the company had never tried before. The agency also said it was “in close contact and collaboration with the United Kingdom, Turks & Caicos Islands, Bahamas, Mexico, and Cuba.”
Coleman told ProPublica that the concerns of the Caribbean countries, along with Starship’s prior failures, helped convince the FAA to close more airspace ahead of Flight 9.
On May 27, the craft lifted off at 7:36 p.m. EDT, an hour later than in March and two hours later than in January. The FAA said it required the launch window to be scheduled during “non-peak transit periods.”
This mission, too, ended in failure.
Starship’s Super Heavy booster blew up over the Gulf of Mexico, where it was supposed to have made what’s called a “hard splashdown.”
In response, the FAA again activated an emergency no-fly zone. Most aircraft had already been rerouted around the closed airspace, but the agency said it diverted one plane and put another in a holding pattern for 24 minutes. The FAA did not provide additional details on the flights.
According to the agency, no debris fell outside the hazard area where the FAA had closed airspace. Pieces from the booster eventually washed up on Mexico’s beaches.
Starship’s upper stage reached the highest planned point in its flight path, but it went into a spin on the way down, blowing up over the Indian Ocean.
The Path Ahead
A map released by the FAA shows potential no-fly zones planned for future Starship launches that would cross over a portion of Florida. Air hazard areas — the AHAs on this map — are paths that would be cleared of air traffic before launches. Federal Aviation Administration
SpaceX launched Starship again in August and October. Unlike the prior flights, both went off without incident, and the company said it was turning its focus to the next generation of Starship to provide “service to Earth orbit, the Moon, Mars, and beyond.”
But about a week later, Transportation Secretary Sean Duffy said he would open up SpaceX’s multibillion-dollar contract for a crewed lunar lander to rival companies. SpaceX is “an amazing company,” he said on CNBC. “The problem is, they’re behind.”
Musk pushed back, saying on X that “SpaceX is moving like lightning compared to the rest of the space industry.” He insulted Duffy, calling him “Sean Dummy” and saying “The personresponsible for America’s space program can’t have a 2 digit IQ.”
The Department of Transportation did not respond to a request for comment or make Duffy available.
In a web post on Oct. 30, SpaceX said it was proposing “a simplified mission architecture and concept of operations” that would “result in a faster return to the Moon while simultaneously improving crew safety.”
SpaceX is now seeking FAA approval to add new trajectories as Starship strives to reach orbit. Under the plan, the rocket would fly over land in Florida and Mexico, as well as the airspace of Cuba, Jamaica and the Cayman Islands, likely disrupting hundreds of flights.
In its letter, the pilots’ union told the FAA that testing Starship “over a densely populated area should not be allowed (given the dubious failure record)” until the craft becomes more reliable. The planned air closures could prove “crippling” for the Central Florida aviation network, it added.
Still, SpaceX is undeterred.
Diez, the company executive, said on X in October, “We are putting in the work to make 2026 an epic year for Starship.”
