Connect with us

Tech

X says it found and took down a Chinese bot farm posting anti-AI data center content to thousands of followers

Published

on


  • An X investigation just found 200k Chinese bot accounts, including 200 posting anti-AI content
  • The posts were targeting and amplifying legitimate US citizen concerns about data centers
  • Is this part of a Chinese effort to slow down American innovation, to race ahead?

X’s Global Government Affairs account has revealed it “identified a bot farm of approximately 200,000 accounts,” 200 of which “posting in a manner that could manipulate a legitimate debate about American AI and energy policy.”

Some of the narratives that these 200 accounts were promoting included that: AI data centers are pushing up household electricity prices, rapid data centre construction is placing excessive strain on the US grid, and ordinary households are subsidising wealthy AI and data center companies.

Source link

Advertisement
Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Tech

Early Leak Of NVIDIA’s DLSS 5 Has An Uncanny Valley Problem

Published

on

People are grafting a pre-release version onto their favorite games with unspectacular results.

Sometimes, photorealism (or something approaching it) makes a video game look better. Other times, well, you get whatever this is from NVIDIA’s leaked build of DLSS 5. The unfinished software, extracted from an early-access PC version of NBA 2K27, has been grafted onto various games. The results are… let’s just say, not universally beloved.

DLSS 5’s neural rendering model takes the upscaling technology in a new direction. Rather than simply reconstructing a sharper version of the image the game rendered, it “infuses pixels with photoreal lighting and materials,” according to NVIDIA. In practice, the results look more uncanny valley than truly next-gen. It’s like someone fed a still image from the game into Midjourney and prompted it to “make it look photorealistic.”

Advertisement

In fairness, the images and videos circulating online are from an early, unofficial build of NVIDIA’s upscaling tech. But with that requisite caveat out of the way, let’s have a laugh at how surreal and off-putting some of these results are.

Kill it… kill it with fire

Take Jesse Faden from Control, one of the first games to get the unofficial DLSS 5 treatment. In the game, she’s supposed to be roughly 28. But after applying a little DLSS 5 magic, she suddenly looks middle-aged. You can kinda see how the AI could get there from the original face. But it still misses the mark, simultaneously looking more real and more fake.

Advertisement

Then we have Skyrim (because, no matter the subject, we’ll always have Skyrim). DLSS 5 makes its lower-poly characters look like a 3D scan of a real person plastered onto a 15-year-old video game character.

Even setting aside how the technology can dramatically alter faces, there’s the fact that the underlying character models and animations are left untouched. That creates a bizarre mismatch that only heightens the uncanny effect. The faces may look kinda real-ish — until you see them stare blankly into the distance with suspiciously stiff movements.

If all of this kept performance in check, it might be strange but harmless. After all, if a developer wants to implement it and a player likes it, knock yourselves out. But that isn’t the case, at least with this leaked build. With Neural Rendering enabled, Control on an RTX 5070 Ti reportedly dropped from 71 fps to 35 fps at 4K. If the final version looks weird and kills performance, it’s hard to see this as anything but NVIDA — now an AI company first and foremost — losing touch with the gaming audience that took it to the top.

DLSS 5 is scheduled to launch in its official form this fall. Once developer-approved versions start rolling out, it’ll be interesting to see whether intentionally integrated uses of the final technology look and feel more natural.

Advertisement

Source link

Continue Reading

Tech

Russia is now shopping a 1,000-kilometre kamikaze drone abroad after putting its long-range strike weapon through live demonstrations

Published

on


  • Russia is offering the Garpia-A1E to foreign military customers
  • The drone can reportedly travel up to 1,000 kilometres
  • Garpia-A1E carries over 50kg payload and endures for over six hours

Moscow has begun marketing a long-range loitering munition built by a state defense manufacturer to foreign militaries.

The weapon, known as the Garpia-A1E, has drawn interest from buyers abroad following a series of live demonstrations.

Source link

Continue Reading

Tech

Yamaha expands its streaming line-up with the WXA-70A and WXC-70A

Published

on

Yamaha has launched the WXA-70A streaming amplifier and WXC-70A streaming pre-amplifier as its newest hi-fi separates.

Both units share an ESS SABRE ES9039Q2M digital-to-analogue converter, a chip more commonly found in higher-priced separates, alongside Yamaha’s proprietary D.O.P.G. circuit design that isolates delicate analogue signals from digital noise.

That same DAC works alongside Yamaha’s ASRC clock-rebuilding technology to reduce jitter before the audio signal reaches the converter, preserving timing accuracy across sources ranging from wireless streaming to TV audio.

The WXA-70A pairs that digital architecture with Class D amplification from ICEPower’s newest 125ASX2 generation, and the amplifier outputs 70 watts per channel into 8 ohms and 105 watts per channel into 4 ohms.

Advertisement

Alongside that amplification, YPAO room calibration uses an included microphone to measure the listening space and adjust playback so the soundstage better suits your room.

Advertisement

The pair can also connect to a television through HDMI eARC or ARC, and a compatible TV remote can control volume and power through that HDMI connection.

Yamaha WXC70A lifestyleYamaha WXC70A lifestyle
Image Credit (Yamaha)

Dialogue Level control adjusts vocal clarity in films and sport without raising overall volume, while MusicCast multi-room support lets the WXA-70A and WXC-70A link with other compatible Yamaha speakers around the home.

Streaming support extends across AirPlay 2, Google Cast, Roon Ready, Spotify Connect, Tidal Connect and Qobuz Connect, a combination that covers most major services without a separate hub or bridge device.

Advertisement

For smart home installation, both models come with Control4 integration alongside a built-in web setup tool, an IR input for wired control systems and 12V trigger support for synchronising power with other connected components.

Advertisement

That level of control extends to wireless connectivity, since installers can disable Wi-Fi and Bluetooth individually during setup to prevent guests or nearby devices from selecting the wrong source by accident.

Pricing for the WXA-70A is set at £1,199, while the WXC-70A is available for £859. Yamaha has confirmed both models will reach UK retailers in November this year.

Advertisement

Source link

Continue Reading

Tech

4 Ireland-based finance internships to boost your CV and career

Published

on

If you are a recent graduate or have time in between your educational and professional lives, consider an internship to boost skills and put you on the right path.

Internship programmes are an ideal way for young people, new graduates and those looking for a career change to add to their skillsets and develop a stronger network of peers and mentors. 

As August is the time when SiliconRepublic.com takes a closer look at careers and skills in the finance and fintech sectors, what better way to close out the month than to take a look at some of the relevant internships open to potential applicants?

Abbott

US medical device and healthcare company Abbott has room for a finance intern at its Dublin facility, supporting the business team. The successful applicant will gain exposure to financial planning, reporting, analysis and process improvement activities. The internship will also include specific assignments across the company. Placements can be engaged with over six, nine or 12 months depending on the needs of the applicant and 19 September is the closing date.  

Advertisement

Carrier

Smart climate and energy solutions company Carrier is in the process of recruiting a motivated and detail-oriented treasury intern, who will join the growing treasury team based in the newly established Carrier Dublin Treasury Centre.

The programme is nine months long and will involve supporting the team in executing strategic projects including in bank account reduction, bank account certification and master data updates. The ideal candidate will either have or be in pursuit of a BA in finance or accounting. Additional certification is favourable but not a requirement. 

Fineco Asset Management

Dublin-based asset management company, Fineco Asset Management, is looking for an intern to join its team. The right applicant will be motivated and detail-oriented and will gain hands-on experience in the financial services industry through working closely with the finance team and other departments in the company. Ideally, whomever earns the position will have a strong analytical mindset and a keen interest in financial markets and investment management.

Microsoft

Tech giant Microsoft has a Dublin-based programme that opens and closes applications between August and October typically. Established in 2003, the Finance Rotation Programme welcomes around 60 interns and 70 graduate students each year to experience financial analysis and is geared towards people with no industry experience.

Advertisement

The programme takes two years, involves four- and six-month rotations and allows students to select rotations based on interests and career aspirations. 

Tips and tricks

If you are interested in taking part in an internship, there are a number of factors you should consider before accepting, to ensure that your placement is the correct one for you. 

One of the most important aspects of an internship is its potential to create a path towards your future career, but that doesn’t mean that organisations hold all the cards and the experience is always worth it without compensation. So before you commit, make sure that – whether it is paid or unpaid – you are not put in a financially precarious position.

If it is unpaid, but the experience is truly worth it, then it might well be the one for you. But if there is the potential that you might leave the role without having developed professionally, while also not having received fair compensation, then it might not have been of real value

Advertisement

You should also consider if the area of finance or fintech you chose to learn more about via an internship has the potential to further your skills. 

It can be tempting to choose a programme because the benefits are great, because it is convenient or because you respect the organisation, but if it doesn’t improve your abilities, or put you into contact with the type of people who can make things happen for you, then arguably it is wasted time. 

But don’t worry, by doing a little research, narrowing down your goals and connecting with people who have forged the path ahead of you, it is possible to determine what is a good fit and what might be wise to pass on. 

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Advertisement

Source link

Continue Reading

Tech

Casting Engine Parts From 3D Prints

Published

on

After building a couple of internal combustion engines by milling billet aluminium stock and cringing at the absolute waste of material this created, [Camden Bowen] figured he’d give casting metal parts a shot. Of course, the key here is to create the molds for said casting, which is where you got a few options available.

Since DIY is really his thing, he also made his own kiln using cement and perlite, plus a propane burner. For the aluminium material to melt, he bought a stack of aluminium alloy wheels, as these are made of an alloy that’s actually suitable for casting. These were turned into ingots as a first step towards casting the engine parts, which among other things helps to purify the metal.

For the actual casting method he picked lost PLA, meaning the intended shape is 3D printed in PLA, then put into plaster before it’s melted out of the newly minted mold in an oven and subsequently burned out in the kiln. For the plaster [Camden] used regular Plaster of Paris, mixed with sand to give it suitable heat-resistant properties.

After some trial and error, as well as a lot of trouble burning out all the PLA, he got a usable mold and managed to eventually cast an engine cylinder with only a few imperfections. Considering just how convoluted it would have been to mill that part out of billet aluminium, it’s easy to see why commercial manufacturers are casting such parts as well.

Advertisement

Source link

Advertisement
Continue Reading

Tech

Brave’s browser one-ups Chrome with its new support for email aliases

Published

on

Chrome alternative Brave just added a good reason to make the switch to its browser: email aliases. The feature, announced this week, allows Brave’s users to sign up for websites and other online services without having to share their personal email addresses.

To use the feature, you first need to create an account with Brave, where you can provide your real email address. Then, once logged into the browser, you’ll see a pop-up appear when clicking an email field on a website, offering the option to use an alias instead. (If the pop-up doesn’t appear, just right-click on the field to access the option.) The email alias will be filled in on the website’s form, and any emails coming to that address will be forwarded to your primary email account.

Image Credits:Brave

The company explains that using an alias can protect your privacy, particularly because websites use your email as a personal identifier, allowing them to target you for ads. Ad tech giants like Meta are also able to match the email you provided to a retailer’s website, for instance, with the address they have on file for you, allowing these companies to track what you bought or viewed.

Plus, if any website where you shared your email gets hacked, the hackers will then know your primary email address, and this information can be leaked and shared with data brokers.

Brave says it built the new feature to plug this privacy hole, adding that it doesn’t read the contents of emails sent to an alias — it only processes them for spam and virus filtering. After the email is forwarded, Brave deletes it from its servers. All data in users’ Brave accounts is also encrypted at rest, and any notes you save alongside an alias are stored locally on your device, unless you turn on the sync option. If syncing is enabled, notes will be encrypted end-to-end.

Advertisement

As the email alias feature is still new, Brave cautioned that some of the forwarded mail may initially go to your spam folder, but this should improve as Brave builds up its reputation score as a mail provider.

For the time being, Brave users are being offered five free email aliases. More will be available for Premium plan users in the future.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Source link

Advertisement
Continue Reading

Tech

Pete Hegseth And His Department Of Unconstitutional Revenge Lose Again In California

Published

on

from the when-the-ai-company-is-more-principled-than-the-government dept

When Pete Hegseth declared Anthropic a national security risk because it wouldn’t agree to his dangerously unethical demands for what Claude should be able to do, Anthropic sued to challenge the directives. One of the challenges was brought in the Northern District of California, where Anthropic has now notched another significant win.

While the previous victory resulted in a preliminary injunction being issued against Hegseth and his Department of Defense (unfortunately the court and Anthropic referred to it by the Trump Administration’s nickname, “Department of War,” but until Congress agrees to the renaming, we won’t), the court has now ruled on the merits, finding in favor of Anthropic on most of its claims and giving it all the relief it sought, namely both a permanent injunction and vacatur of Hegseth’s agency action.

In short, the court found that the action to “designate Anthropic a supply chain risk to national security, order all federal agencies to permanently stop using Anthropic’s products, and ban all defense contractors from doing any business with Anthropic, even if unrelated to the military” violated Anthropic’s First Amendment rights, as well as its due process rights, and it further violated the APA as an arbitrary and capricious action.

The undisputed record shows that the challenged actions constituted unlawful retaliation in violation of the First Amendment, and that Anthropic was denied the pre-deprivation process required under the Fifth Amendment. The record further shows no material dispute that Secretary Hegseth’s decision to designate Anthropic a supply chain risk violated the governing statutory scheme, 10 U.S.C. § 3252, and was arbitrary and capricious. Though the Department of War is undisputedly free to select the AI vendor of its choice, the evidence demonstrates that the broad measures imposed on Anthropic were illegal and baseless.

On the First Amendment claims, “The undisputed facts establish that Anthropic’s protected speech, on a matter of great public importance, was a substantial motivating factor for Defendants’ speech-chilling conduct, and that Defendants would not have taken the retaliatory action absent their desire to make an example of Anthropic for its public stance on the weighty issues at stake in the contracting dispute.” In reaching this conclusion the court found that Anthropic had been engaging in speech on matters of public concern—such as the safeguards built into an AI system—and that the Administration’s own social media posts made it clear that designating Anthropic as a national security risk was designed to punish it for this very expression articulating why it did not want to sell the government a version of Claude without those safeguards, rather than for any legitimate purpose.

Advertisement

Secretary Hegseth expressly tied Anthropic’s punishment to its attitude and rhetoric in the press. He stated that “Anthropic delivered a master class in arrogance.” (AR 255B.) Referring to Anthropic and Amodei, he further stated: “Cloaked in the sanctimonious rhetoric of ‘effective altruism,’ they have attempted to strong-arm the United States military” through their “corporate virtue-signaling” and “Silicon Valley ideology.” (Id.) “Anthropic’s stance is fundamentally incompatible with American principles.” (Id.) The President described Anthropic as a “RADICAL LEFT, WOKE COMPANY” and its employees as “Leftwing nut jobs,” who “made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War.” (AR 255A.) […] Furthermore, the decision to publicly broadcast Anthropic’s punishment via social media—even before the formal, administrative process of making the necessary findings to designate Anthropic a supply chain risk had begun—makes little sense except as an attempt to swiftly make a public example of Anthropic for daring to criticize the Administration.

In other words, the government did not have to buy Claude if it couldn’t agree with Anthropic on what it should be able to do, but designating it a supply chain risk ineligible for any government contract was a punishment inflicted because it did not like Anthropic’s viewpoint that AI should have the safeguards Hegseth did not want it to have.

On its First Amendment claim the court found that Anthropic had shown “that (1) it engaged in constitutionally protected activity; (2) the defendant’s actions would ‘chill a person of ordinary firmness’ from continuing to engage in the protected activity; and (3) the protected activity was a substantial motivating factor in the defendant’s conduct.” It decided that the alternative Pickering test did not apply (“The Pickering framework is inapplicable here because the Challenged Actions were an exercise of sovereign power, not contractual power. […] Furthermore, the public labeling of Anthropic as an adversary of the United States and an “out-of-control, Radical Left AI company” that was attempting to “seize veto power over the operational decisions of the United States military” reflects a concerted effort to make an example of Anthropic” for its viewpoint resisting government demands, and not part of ordinary contract negotiations.), but even if it did, Anthropic would have still prevailed.

Moreover, even if the more flexible Pickering framework were applied, Anthropic has still shown that it is entitled to summary judgment. Under the framework, Anthropic must show it engaged in expressive conduct about a matter of public concern; government officials took adverse action against it; and its expressive conduct was a substantial or motivating factor for the adverse action. Damiano v. Grants Pass Sch. Dist. No. 7, 140 F.4th 1117, 1137 (9th Cir. 2025). In other words, Anthropic’s prima facie showing under the traditional test satisfies its burden under the Pickering framework so long as Anthropic is speaking on a matter of public concern. As already explained, Anthropic satisfies this burden. The government argues that “Anthropic’s public airing of its objections to DoW’s contractual terms does not transform this into a matter of public concern protected by the First Amendment.” (Dkt. No. 214 at 10.) However, “attempt[s] to reach the general public” are “considered relevant in other public concern cases.” Havekost v. U.S. Dep’t of Navy, 925 F.2d 316, 319 (9th Cir. 1991). And as the Supreme Court explains in Janus, matters of “great public concern” can arise in the context of contract negotiations, and the government does not get a pass on upholding First Amendment protections simply because it is engaged in contract negotiations. Janus, 585 U.S. 878, 910–14 (“[I]t is impossible to argue that the level of . . . state spending for employee benefits,” the subject of the contract negotiations between the union and the government, “is not a matter of great public concern.”). Anthropic was not critiquing a “single contract” (Dkt. No. 238 at 8); it was discussing the appropriate limits on the government’s use of frontier AI technology. That is a matter of public concern.

And the government had nothing to rebut Anthropic’s case, including any sort of true national security argument.

Defendants’ asserted national security concerns are further contradicted by their actions shortly before, immediately after, and since the Supply Chain Designation. Shortly before, in the February 24 meeting, Secretary Hegseth raised the possibility of invoking the Defense Production Act, meaning Anthropic was essential to national security rather than a threat to it. (Dkt. No. 166-5 ¶ 17; Dkt. No. 166-9 at 3.) Immediately after the designation, Under Secretary Michael told Anthropic that a deal was “very close.” (Dkt. No. 166-5 ¶ 27.) Since the designation, the White House has “discussed opportunities for collaboration” with Anthropic utilizing its new Mythos model, and these discussions appeared to range across a variety of sensitive contexts. (Dkt. No. 166-4 ¶ 81.) Those actions are inconsistent with any genuine belief that Anthropic is an adversary of the United States who might secretly poison its model to undermine national security. As such, the government has failed, under even a deferential review, to show that a legitimate national security interest outweighed Anthropic’s speech rights.

The court also found that the government’s action violated Anthropic’s due process rights. “Due process requires notice ‘reasonably calculated, under all the circumstances, to apprise interested parties of the pendency of [a government] action and afford them an opportunity to present their objections,’” but here Anthropic had none before it was suddenly blacklisted and designated a supply chain risk. That action deprived it of a liberty interest unconstitutionally.

Advertisement

“[A]bsent exigent circumstances,” due process must happen “before [a] final deprivation.” Esparraguera v. Dep’t of the Army, 101 F.4th 28, 40 (D.C. Cir. 2024). The record reflects that the Challenged Actions were taken without any meaningful notice or pre-deprivation process (and, in the case of the Presidential Directive and the Hegseth Directive, without offering any post-deprivation agency process either). Although Anthropic was on notice that the government objected to its contracting terms, it had no notice or opportunity to object before Defendants publicly barred it from all federal government work and blacklisted it with defense contractors. It also had no notice or opportunity to object to the factual basis for its designation as a supply chain risk, which it learned of in this litigation.

And then, in addition to the constitutional violations, the court also found that the designation was an arbitrary and capricious act under the Administrative Procedure Act as well.

Under the APA, an agency action must be set aside and held unlawful if it is “arbitrary, capricious, an abuse of discretion,” “in excess of statutory jurisdiction, authority, or limitations, or short of statutory right,” or “without observance of procedure required by law.” 5 U.S.C. § 706(2)(A), (C), (D). Anthropic has shown that the Hegseth Directive and the Supply Chain Designation were in excess of statutory authority, contrary to law, and arbitrary and capricious. […] Secretary Hegseth’s February 27 order that “[e]ffective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic” is a final agency action and thus is subject to APA review. (AR 255B.) On its face, the order is the “consummation” of a decision-making process and “legal consequences [] flow” from it. Bennett v. Spear, 520 U.S. 154, 178 (1997) (citations omitted). Absent recission, which has not occurred to this day, any company doing business with both Anthropic and DoW between the time of the post and when the preliminary injunction issued in this case would have been in violation of the order. Indeed, Anthropic submits evidence that several law firms sent client alerts “describing the potentially far-reaching nature of the government’s actions and suggesting that Department contractors may be best served by reevaluating their relationship with Anthropic.”

The court didn’t buy the government’s defense that “Secretary Hegseth could not have meant what he said because he concededly lacked the requisite authority to issue such a broad order.” To do so would have eviscerated the APA and its prohibition against arbitrary and capricious acts.

After all, if courts were to find agency actions unreviewable as non-final simply because the agency had not met its statutory obligations or did not have authority to take the action, that would defeat the entire purpose of APA contrary to law review. There is no way to read Secretary Hegseth’s words as anything other than a “consummation” of a decision-making process. (See Dkt. No. 238 at 17 (conceding in Defendants’ reply brief that “[w]hatever else, there is no doubt about the Secretary’s commitment to his course of action”).) And his Directive imposed “legal consequences” by declaring, effective immediately and in unqualified terms, that companies could not continue doing business with both Anthropic and DoW, with the implicit threat that DoW would stop contracting with violators. See San Francisco Herring Ass’n v. Dep’t of the Interior, 946 F.3d 564, 580 (9th Cir. 2019) (finding legal consequences flowed from Park Service orders where there was “no suggestion that compliance . . . was somehow optional”).

Furthermore, Anthropic did not meet the statutory definition for “supply chain risk” anyway.

The Administrative Record establishes that Anthropic’s conduct does not meet the requirements for finding a “supply chain risk” under Section 3252. The Michael Memo states: “By embedding unreasonably restrictive terms that restrict DoW’s warfighting operations beyond the limitations imposed by law, Anthropic seeks to grant itself an operational veto. This triggers the legal definition of supply chain risk at 10 U.S.C. § 3252(d)(4) . . . .” (AR 213.) But Section 3252 defines a supply chain risk as limited to “the risk that an adversary may sabotage, maliciously introduce unwanted function, or otherwise subvert . . . a covered system.” 10 U.S.C. § 3252(d)(4). The plain text of the statute is directed at covert acts or hacks, not overt, public positions regarding contract terms. The legislative history similarly explains that the statute aims to address “increasing risk that systems and networks critical to [the Department of Defense] could be exploited through the introduction of counterfeit or malicious code and other defects introduced by suppliers of systems or components.” S. Rep. No. 111-201, at 162 (2010). Contrary to Under Secretary Michael’s conclusion, Anthropic’s contracting position does not appear to bear any relation to the conduct that Section 3252 is aimed at addressing.

And per the other statute in play, it also did not grant Hegseth the authority to do any of this either.

Advertisement

Agencies may not impose sanctions or issue orders “except within jurisdiction delegated to the agency and as authorized by law.” 5 U.S.C. § 558(b). The Court has already found that the Presidential Directive to “EVERY Federal Agency in the United States Government to IMMEDIATELY CEASE all use of Anthropic’s technology” was a retaliatory act, taken without due process, meant to punish Anthropic for its protected speech. Supra § V.A–B. The record shows that the following Agency Defendants issued orders to terminate use of Anthropic’s products pursuant to the Presidential Directive, most within hours of the directive being issued: Treasury, FHFA, State, GSA, OPM, NRC, DHS, and Energy. Supra § II.G–H. DoW did the same, in reliance on the Hegseth Directive and Supply Chain Designation. Supra § II.E. These orders, in violation of Anthropic’s First Amendment and due process rights, were not “authorized by law,” and violate Section 558(b).

The court did not, however, find in favor of Anthropic’s ultra vires claim, because it hinged more on Trump, who was not named in this case, and seemed to have been underpled, compared to the other claims, but ultimately it was not necessary for Anthropic to prevail on to get all the relief it sought.

When a presidential order does not stem from powers granted through “an act of Congress or from the Constitution itself,” the order violates the separation of powers doctrine, and a district court may enjoin federal agencies from implementing the order as ultra vires. Youngstown Sheet & Tube Co. v. Sawyer, 343 U.S. 579, 584–86 (1952). Anthropic argues that the president lacks “statutory authority” or a “constitutional basis” to direct “EVERY Federal Agency in the United States Government to IMMEDIATELY CEASE all use of Anthropic’s technology.” (Dkt. No. 166 at 30; AR 255A.) In passing, Anthropic cites four statutes regulating “exclusion from federal contracting.” (Dkt. No. 166 at 30 (citing as “examples” 10 U.S.C. §§ 3203(a)(1), 3204(a); 41 U.S.C. §§ 3303(a)(1), 3304(a)).) However, Anthropic is not asserting a claim for violation of a specific procurement statute, see, e.g., State v. Su, 121 F.4th 1, 5 (9th Cir. 2024), and “claims simply alleging that the President has exceeded his statutory authority are not ‘constitutional’ claims.” Dalton v. Specter, 511 U.S. 462, 473 (1994). Anthropic’s passing references to the statutory provisions and Congress’s “power of the purse” are insufficient to carry its burden of showing the President clearly exceeded “any background constitutional authority” to issue the Presidential Directive, thereby rendering the order ultra vires in violation of the separation of powers. Sierra Club v. Trump, 929 F.3d 670, 696 (9th Cir. 2019).

That the court found liability for both the constitutional claims and APA claims is why Anthropic was able to get all that relief, including both vacatur of the agency action and also a permanent injunction.

First, despite the government’s arguing that the matter should simply be remanded back to the agency for further action, rather than the court vacating what the agency had earlier done entirely, the court disagreed that such a remand would be adequate (“In light of Anthropic’s showing that the Supply Chain Designation violates the substance of Section 3252 and that Secretary Hegseth lacked any authority to order a secondary boycott, remand without vacatur would be an inadequate remedy.”). There was no point giving the government another bite of an apple it wasn’t entitled to eat in the first place.

Nor were the government’s claims it would be harmed by this more extreme remedy availing, given how the government had been coping just fine with its action already having been enjoined for quite some time (“[T]here has been no showing by the government that vacatur would result in a national security risk or any other harm—despite the fact that the Supply Chain Designation and the Hegseth Directive have been preliminarily enjoined for over five months.”). It is also why the court denied an automatic stay of its order, since the government had obviously not been harmed by the preliminary injunction already in place.

Advertisement

And as for why the court also granted a permanent injunction on top of vacatur, it found that vacatur alone would not redress the constitutional injury.

Furthermore, to the extent Defendants rely on their voluntary cessation of the challenged conduct, it is not “absolutely clear that the allegedly wrongful behavior could not reasonably be expected to recur.” Friends of the Earth, Inc. v. Laidlaw Env’t Servs., 528 U.S. 167, 189 (2000) (citation omitted); Enrico’s, Inc. v. Rice, 730 F.2d 1250, 1253 (9th Cir. 1984) (“We recognize that the voluntary cessation of allegedly illegal conduct does not deprive a court of the power to grant injunctive relief.”). Defendants continue to defend and have declined to stipulate to enjoin the Challenged Actions. For these same reasons, legal remedies would be inadequate. As to Defendants’ argument that Anthropic “continues to speak freely,” has experienced an increase in its valuation, and has not had its federal contracts terminated yet (Dkt. No. 214 at 33–34), that reflects the restoration of the status quo pursuant to the preliminary injunction in this case. The record shows no indication that, upon lifting the preliminary injunction, the harms at issue would still be averted.

In sum, the vindictive petulance of Hegseth and the Trump Administration has led to a particularly obvious constitutional injury, which no statute otherwise privileges. Per the court, the directives are now nullities and Hegseth and his agency are prohibited from trying again. Pursuant to the separate order, if they want to take any action, it can only be what they were lawfully allowed to do in the first place.

This Order does not bar any Defendant from taking any lawful action that would have been available to it on February 27, 2026, prior to the issuance of the Challenged Actions. For example, this Order does not require the Department of War to use Anthropic’s products or services and does not prevent the Department of War from transitioning to other artificial intelligence providers, so long as those actions are consistent with applicable regulations, statutes, and constitutional provisions.

Filed Under: 1st amendment, administrative procedure act, dario amodei, defense department, dos, free speech, pete hegseth, supply chain, supply chain risk

Companies: anthropic

Advertisement

Source link

Continue Reading

Tech

How Long Is A New Fire TV Stick Actually Supposed To Last?

Published

on

A Fire TV Stick can outlive its guaranteed support window, though aging apps and hardware may make you want to replace it sooner.

There isn’t an expiration date hidden in the settings menu of a Fire TV Stick. Amazon also doesn’t reveal a typical physical lifespan for its streaming sticks, nor does it inform owners to replace one after three, four or five years. That makes the question of “how long will it last?” a bit more complicated than it sounds.

At least one of the numbers is firm. Amazon says security updates are provided for at least four years from the date the Fire TV device was purchased new from Amazon as the seller, or until the support date listed for that model, whichever is later. The Fire TV Stick 4K Select and Fire TV Stick 4K Plus are among the current models with security updates guaranteed through at least Dec. 31, 2030.

Amazon doesn’t call it a lifespan, though. A support deadline indicates the duration of security fixes, not when the hardware will decide it has streamed its last episode.

Advertisement

Four years is a minimum, but it’s not a countdown to the junk drawer.

Four years isn’t an expiration date

The Fire TV Stick can be fully functional beyond the security-update window, but simply turning on isn’t a very high standard for a streaming device. Older hardware can become increasingly uncomfortable to use as apps become more demanding, interfaces evolve and video standards advance, even if nothing has failed internally.

Video codecs are a part of that. Newer video codecs like H.265/HEVC and AV1 can compress video more efficiently than older codecs, but they also need the right decoding hardware. The Plus doesn’t have an advantage on that front, as Amazon’s current Fire TV Stick 4K Select and 4K Plus both offer hardware decoding for AV1 and H.265 at up to 4K60.

Advertisement

Apps can be less accommodating. Netflix itself notes that its app may no longer be available on some TVs and streaming devices made before 2015, providing a neat example of what aging really looks like for a streamer: The hardware can still work, while one of the reasons you bought it quietly stops working with it.

That doesn’t mean every Fire TV Stick has a 10-year lifespan, but it does show why declaring one “dead” simply because four years have passed is difficult to justify.

Advertisement

When should you actually replace your Fire TV Stick?

Slower streaming performance isn’t necessarily a retirement notice, either. Before giving up on your Fire TV Stick, Amazon suggests restarting it, installing available updates, clearing app caches and ruling out network issues. Insufficient power can also cause restarts, making a healthy stick look much more dramatic than it is.

Persistent problems are more convincing. Apps crashing repeatedly, services dropping support for your model, menus staying slow after troubleshooting or storage becoming a constant hassle are signs that an upgrade may actually help.

Buying slightly higher up the range can give you more breathing room. The Fire TV Stick 4K Select has 1GB of RAM and costs about $40, while the 4K Plus has 2GB for about $50. Amazon warns that memory pressure can kill apps on lower-memory Fire TV hardware, so that extra $10 buys more headroom as software gets heavier.

It doesn’t guarantee an extra year or two, and Amazon makes no such claim. The Select’s Vega OS has an efficient footprint and was designed to scale across devices. Still, doubling the RAM for roughly the price of lunch is worth considering if you’ll keep the stick for years.

And then there’s the remote. Its two AAA batteries will need replacing, and a failed remote can be replaced separately. Amazon’s Fire TV mobile app can control the streamer too, so a dead remote doesn’t necessarily mean a dead Fire TV Stick.

Advertisement

Netflix discontinued support for first-generation Fire TV and Fire TV Stick devices in June 2025, The Verge reported, more than 10 years after they launched in 2014.

Source link

Advertisement
Continue Reading

Tech

Microsoft Teams Has Become a Haven for Scammers in China

Published

on

On Chinese social media, people have also reported experiencing variations of the scam involving other corporate software apps, including Webex, a video conferencing platform owned by Cisco, and Cliq, a workplace communication app owned by the Indian software giant Zoho. Since February 2025, 71 percent of the over 150 reviews of Webex on Apple’s Chinese app store referenced being scammed on the platform, according to WIRED’s analysis. Cisco did not respond to a request for comment.

Sam Wunderl, a spokesperson for Zoho, says the company has identified “a limited number of instances involving scammers using Zoho Cliq to defraud victims.” The company says it uncovered the suspicious usage internally. As of August 27, Zoho says it disabled online payments to Cliq in China, suspended every account created by the suspected scammer it identified, and plans to discontinue the free version of Cliq in China going forward.

All Roads Go to Teams

According to dozens of victim accounts shared on Chinese social media platforms like Xiaohongshu and Douyin, the scammers look for new targets seemingly anywhere they can. They pose as job recruiters looking to fill open positions, prospective renters contacting landlords, potential dates on Tinder, and international buyers looking to source products from Chinese factories.

The scammers take advantage of Teams’ enterprise features by creating accounts for their targets and sending them pre-made login credentials. Because the scammers control the organization, they can deactivate the accounts once the scheme is over, wiping out victims’ access to chat histories that could be potentially used as evidence. To explain the unusual setup, scammers may tell victims they’re using work devices that only allow certain apps, or say they created a private Teams account just for the two of them to chat.

Advertisement

It turned out that some of the other victims who contacted Zhao had encountered the very same scammer she had. One was able to share Teams login credentials with her. This time, Zhao pretended to be a new potential victim in order to hopefully trick the scammer into providing more information about how his ruse worked.

She asked why he had to use Teams instead of WeChat, by far the most popular messaging app in China. The scammer claimed his colleagues could see his WeChat messages because he was on a work project, so Teams “is like our secret base together.” The account he gave Zhao, he said, was “for team members to contact family while on the project.”

“Can I ask a small question. What’s the special meaning about my account ‘joy20706905?’” Zhao asked, referring to the username he provided. “The account was assigned by the software when you purchase it from Microsoft,” he responded.

Zhao says when she was defrauded in May, Microsoft Teams didn’t alert her that the conversation she was having with the scammer might be dangerous. It was only when she talked to him again in August that the app started showing a banner that reads: “Don’t share confidential info or your screen with people you don’t know. Report anything suspicious.”

Advertisement

Why Teams?

In many ways, Microsoft Teams is an ideal app for fraudsters in China. Unlike Telegram or WhatsApp, it isn’t blocked by the government. And it has advanced tools like screen sharing and remote control that allow scammers to potentially gain more control over their victims’ devices. Most importantly, it’s a legitimate workplace communication app made by a trusted multinational tech company.

The same factors also apply to Webex. Tan Chenxin, a Chinese software engineer living in New York, says he was nearly defrauded earlier this week after he received a call from someone claiming to be working on a joint police investigation with US Customs and Border Protection and Chinese law enforcement. They asked Tan to download Webex and hop on a video call to discuss the situation. “I looked up Webex and found that it was developed by Cisco, a world-leading cybersecurity software provider,” Tan says. “I happen to know what Cisco is, so I convinced myself it was fine.”

Source link

Advertisement
Continue Reading

Tech

Steve Jobs’ 8th-grade science project sells for $34,375

Published

on

Someone did, in fact, buy the Steve Jobs 8th-grade science fair project for roughly the same price as a brand-new midsize sedan.

Earlier in August, I’d reported on yet another Steve Jobs-based auction. These seem to be happening with increased frequency, and I suspect it has something to do with Jobs’ stepbrother, John Chovanec, cleaning out the family home.

Among the usual collection of computer parts, Apple-1 units, and some sealed first-generation iPhones, there were more personal effects. One, of course, being Jobs’ 8th grade science project, which sold for $34,375.

The Apple logo sun catcher, which was my personal favorite, commanded a respectable $4,365.

Advertisement

The biggest ticket item of the auction was an Apple-1 dubbed the “Neumark” Apple-1. It featured a unique Smith-Corona Typewriter case and sold for nearly a half million dollars, coming in at $499,363 by auction’s end.

A “stealth iPod” development prototype nearly doubled its $50,000 estimate, bringing in $91,479 by the time the auction ended.

Other notable items included:

Whoever wound up with the lot of vintage employee apparel got a deal, though. It sold for $875, below the $1,000 estimate.

Advertisement

Unsurprisingly, the Twitter flotsam largely came in barely above the estimates.

Source link

Advertisement
Continue Reading

Trending

Copyright © 2025