And of course, Henry Cavill is involved.

Snap CEO Evan Spiegel sidestepped investors’ questions about preorder demand for the company’s long-awaited Specs smart glasses during Monday’s earnings call, just weeks before the device’s September launch event.

“What we’re hearing from folks is really that they want to try Specs,” Spiegel told investors. “It’s obviously a high consideration purchase at $2,195. Obviously, developers and folks who are familiar with the platform really understand it and understand the technical leaps we’ve made with this generation. I think for the broader public and consumers, it’s going to be really important for folks to go hands-on. Our upcoming launch event will be an important sort of starting point for that consumer-oriented journey.”

The company unveiled Specs in June after spending more than a decade developing the device. The wearable’s $2,195 price tag is significantly higher than most Meta Ray-Ban smart glasses, which start at around $350, but lower than Apple’s Vision Pro, which starts at $3,500.

Investors also pressed Spiegel on why he believes Snap’s strategy is financially viable for a company of its size, why it chose to go it alone rather than partner with another company, and what gives him confidence that the company can compete with Apple, Meta, and Alphabet.

Spiegel responded that Snap believes the long-term opportunity to develop the next computing platform is “enormous.”

“I think what some folks maybe don’t understand yet, especially because Specs are so new and we’re really the first mover in this category, is how difficult the product is to execute from a technical perspective,” Spiegel said. “When we started innovating in the social space, we were a late entrant. So, most of the apps at the time, whether it was Facebook or Instagram or Twitter, were already in existence, and we had to really innovate to continue to grow. What’s so unique about this opportunity for us is really that we’re a first mover, and that really plays to our strengths as an innovator.”

When asked about product-market fit, Spiegel said it will likely be closer to the end of the decade before the company sees mass-market consumer adoption.

“I think things, for example, like weight and cost are going to have to come down to see you know unit volumes really meaningfully pick up.” But we do have, I think, a real advantage here in that developers have been building on the Specs platform now for several years.”

If you’ve ever wanted to copy something on your iPhone and paste it on your Windows PC, the feature is coming in a fall 2027 release of iOS 28 to users in the EU.

Third-party developers can request interoperability features to be developed by Apple via a web portal. If the request is deemed possible it can take anywhere from six months to eighteen months to build the feature and submit it for release.

One such request was made by Microsoft (Request ID: FB22329535) for cross-device clipboard sharing access between iOS and Windows. The request was submitted on March 25, 2026, and after being reviewed by Apple, was approved on June 26, 2026.

Microsoft’s request argued that consumers expect features like clipboard sharing between devices to “just work.” Currently, the solution requires third-party apps to connect the devices, the iOS app to be open in the foreground, and a permission request for each cliard share.

The request suggests that if a solution is developed, it would allow copying from iPhone and pasting to Windows, or vice versa, without needing complicated apps or permission structures. Microsoft shared that it expected the feature would maintain user control and privacy.

Apple’s system shows that the request was picked up on March 31, and after an initial assessment, moved up for further evaluation on April 27. It is interesting to observe Apple’s systems at work, as this goes beyond simple app review or other systems.

A peek behind the DMA curtain

The EU DMA demands that Apple allow developers to request interoperability, and Apple has made it mostly transparent for anyone to view the goings-on. Some requests show as “Confidential,” while others, like Microsoft, show the developer’s name.

Since Apple is fundamentally altering how its OS operates, it is expected for any request to take some time. The clipboard sharing feature seems to fall into the longest-expected timeline of around 18 months.

While Apple says in these submissions that the features will be built for the EU only, there is always the chance they could be released globally. Emulation, for example, was brought about by DMA complaints, but was enabled worldwide.

If Apple builds the shared clipboard system for Windows, it could always make its way to US users, especially if regulators push for it.

iPhone to Windows clipboard sharing

Apple says the solution it has come up with is similar to the Accessory Notifications and Accessory Live Activities frameworks introduced in iOS 26.5 for the EU. Developers will need to implement AccessorySetupKit to enable a one-time permission prompt for users to pair an accessory and share the clipboard.

Of course, the timeline and implementation are Apple’s decision. Whether or not the end result, or the timeline, satisfies Microsoft and the EU is another question entirely.

While Apple didn’t say it explicitly, the feature will be ready for a version of iOS 28. Since it explicitly said “fall” 2027 for the solution’s arrival, it may not be available in the point zero release, which begins beta testing in June.

from the how-is-this-real-life? dept

Sometimes it helps to just write out everything that happened leading up to a situation so we don’t lose sight of just how ridiculous it is.

I mean, would you believe that a President would sue his own IRS for something that happened under his own watch and did him no harm, demand $10 billion, and when a judge calls into question how that’s possible, come up with a “settlement” between the two “parties” that grants himself, his family, and his far-flung businesses (already under investigation for tax fraud) a complete get-out-of-jail-free card from tax audits along with a special $1.776 billion slush fund which he can hand out to insurrectionists who tried to overturn an election he had lost?

And then, his Acting Attorney General seeking the permanent job (who had been the President’s personal lawyer in early criminal cases, including one where the President was convicted of 34 felonies), would tell inquiring Senators that the slush fund wasn’t moving forward, but refuse to put that in writing?

When you write it all out that way, it sounds kinda crazy.

Last month Judge Kathleen Williams nuked the fund and alerted various state bars to the ethical lapses of all the government lawyers involved, including Todd Blanche, the President’s personal criminal defense lawyer-turned Acting Attorney General-turned nominee for the permanent Attorney General position.

Blanche’s nomination had stalled out, in part because two Republican Senators (Cornyn and Tillis, both of whom are leaving office after Trump refused to support their attempts to stay) have questioned the slush fund. Blanche refused to put it in writing that the fund was going away, leading to this strained exchange with Cornyn:

Republican Sen. John Cornyn of Texas, reading from the settlement on Wednesday, noted that it says the terms of the deal “may be modified only upon the written agreement of the parties.”

“Has there been a written agreement of the parties to modify the settlement fund?” he asked.

“No, the settlement fund is just not moving forward,” Blanche responded. “There’s no modification. It never started. No money went from the Treasury to any other account.”

Advertisement

When pressed, however, as to whether the settlement agreement may be enforceable by the parties, Blanche conceded that the president could challenge it down the road.

“Yes, it’s an enforceable document, so I suppose if President Trump’s counsel sought to enforce it, that they potentially could … try to enforce the contract. They can’t force the Department of Justice to move forward with the weaponization fund. They could potentially say that we breached by not moving forward,” Blanche said. “They haven’t done that, and I’m not aware that they’re planning on doing that.”

Given all of that, you would think that, perhaps, Trump wouldn’t make any public effort to say that the fund should still exist.

But, again, we live in the upside down world, where if Donald Trump wants something, he just throws a temper tantrum until he gets it. Three new developments have happened regarding all this in the past few days.

First, on Friday, Donald Trump appealed Williams’ order and filed a long, rambling motion about how unfair the order was. That doesn’t sound like someone who is willing to accept that the fund is dead. Second, on Saturday, Trump posted a rant to his personal social media site about how mad he was at Cornyn and Tillis, and how he still needed the fund, and even hoped Congress would pass a separate law giving him the slush fund. Finally, despite these two things making it abundantly clear that Trump still plans to create a slush fund for his cosplay militia, on Sunday, Blanche finally signed a letter claiming the slush fund was dead.

Let’s take these each in order.

First the appeal and the whiny motion about it. The most incredible thing about it is that, even though it was filed by lawyers hired by Donald Trump (the person), as I was reading it, it read like it was filed by the Justice Department. Remember, the entire reason that Williams killed the “settlement” of Trump’s lawsuit against his own IRS was that there appeared to be no adversarial parties in the case, and it was just Donald Trump suing himself in order to give the “settlement” the stamp of authenticity that it was part of a federal case.

While, to their credit, this new filing at least brings in real lawyers from an actual serious law firm, DLA Piper, alongside Trump’s original two-bit lawyer Alejandro Brito, the argument itself is still a mess. Again, you could totally picture either Donald Trump or his DOJ/IRS filing this since the original ruling applies to both. Because they’re not adverse parties. They’re all on the same side.

The ruling rests on a grievous legal error: that, because President Trump supervises the Executive Branch, he and the United States necessarily possess the same legal interest and cannot be adverse to each other. That is wrong. President Trump asserted a personal claim arising from the theft of his own tax information. The IRS and the U.S. Department of the Treasury (“Treasury”) represented the sovereign’s interests in public funds and federal law. Donald Trump Jr., Eric Trump, and The Trump Organization asserted their own independent claims. Presidential supervision did not erase those rights or merge those interests.

The Sanctions Order nevertheless erroneously treated its novel theory of adversity as the starting point and then refracted the facts through that mistaken lens. Settlement became proof of collusion, even though there was none, potential (weak) defenses became proof that the claims were fictitious, which they were not, and ordinary professional relationships became proof of coordination, which did not exist. But the Sanctions Order identified no pre-filing sub rosa agreement, no false allegation, and no specific direction by President Trump controlling Defendants’ litigation decisions. The predicate legal error thus supplied the conclusion, causing the Sanctions Order to incorrectly recast entirely innocuous facts as misconduct.

The danger posed by the Sanctions Order is profound. The Court called the Article III question “unprecedented,” appointed six amici, and devoted nearly thirty pages to resolving it— yet it then declared the contrary position “so obvious and so insurmountable” that advancing it warranted career-altering sanctions. Id. at 9-38, 53 n.69. That mistake converts disagreement into professional punishment, and strikes at the integrity of the adjudicative process itself. A system that sanctions lawyers for advancing positions on constitutional questions cannot sustain fearless advocacy or principled legal development.

Note the tell in that last bit: the “professional punishment” landed on both sides — Trump’s lawyers and the government’s — which is exactly the point Williams was making. And, yes, courts are historically reluctant to sanction lawyers at all. But that reluctance is the problem, not evidence that the one judge willing to do it got it wrong.

As for the claims that the DOJ was actually representing the IRS’s best interests, and not Donald Trump’s, that’s belied by quotes from Donald Trump himself — who publicly admitted he was negotiating with himself — and the terms of the “settlement” which make no sense and do nothing to support the interests of the IRS, the public, or the American taxpayer. If this case was actually about Trump’s leaked tax returns, how does giving the family a release from tax audits or handing over nearly $2 billion to insurrectionists (wholly unrelated to this suit) have any of the American people’s interests in mind?

The strongest argument this filing has is that one part of the ruling is arguably prior restraint. To make sure that the settlement agreement for the slush fund was not used for improper enrichment, Judge Williams barred the parties from invoking the settlement in other proceedings. Trump’s new lawyers say that this part is prior restraint:

In addition, Rule 11 and inherent authority cannot support an order forbidding the President, private Plaintiffs, federal agencies, and sweeping categories of associated persons from “referring to” the Settlement Agreement or invoking it in any future judicial, administrative, regulatory, arbitral, or other official proceeding. D.E.106 at 47. That unprecedented injunction regulates future speech and advocacy, is unconstitutional, and is patently unlawful.

The Sanctions Order’s command is a content-based prior restraint. It singles out one subject—the Settlement Agreement—and suppresses one message, i.e., that the Settlement Agreement exists and may carry legal effect. Prior restraints carry a “heavy presumption” of constitutional invalidity, Bantam Books, Inc. v. Sullivan, 372 U.S. 58, 70 (1963), and constitute “the most serious and the least tolerable infringement on First Amendment rights.” Nebraska Press Ass’n v. Stuart, 427 U.S. 539, 559 (1976). They require findings that the threatened harm is “both great and certain and cannot be mitigated by less intrusive measures.” CBS, Inc. v. Davis, 510 U.S. 1315, 1317 (1994) (Blackmun, J., in chambers). The Sanctions Order identified no threatened unlawful speech, no resulting harm, and no reason narrower relief would be inadequate. Indeed, it did not address the First Amendment at all.

And while this is the strongest argument, that doesn’t make it actually strong. The order does not stop Trump or his lawyers from talking about the settlement. They can post about it on Truth Social, complain about it on Fox, write a book about it. What they can’t do is walk into another court, agency, or arbitration and invoke a settlement that a federal judge just voided due to the lack of adverse parties. The only “speech” being restrained is further fraud on the court.

But the fact that Trump bothered to make this argument at all is the real tell. You don’t fight for the right to invoke a settlement in future proceedings unless you’re planning to invoke it in future proceedings — which is precisely what Blanche has spent months assuring senators would never happen.

Then, even as Blanche was negotiating with Cornyn and Tillis to get them to vote in favor of his AG nomination by promising not to create the fund, on Saturday, Trump said hell yes he intends to make use of the fund — and added that he’d like Congress to pass a law establishing the same fund as a backup, in case the courts kill the one he negotiated with himself:

That’s a Trump post on Truth Social saying:

If Senators Cornyn and Tillis, both upset because I wouldn’t Endorse them (they lost, and quit, respectively!), aren’t going to approve Todd Blanche, one of the most respected professionals, according to everybody, in the Country, to be the United States Attorney General, then I will keep Todd as Acting A.G., and push hard to get the Anti-Weaponization Bill, which takes care of those who have been so badly treated by the Crooked Joe Biden (and Obuma!) Administration (I get nothing, although I was treated horribly!), PASSED. Todd Blanche was a voice of reason! It will immediately be back on the table, and I will get it done. Thank you for your attention to this matter! President DJT

Once again, Trump’s ability to shoot his allies in the back remains undefeated. Cornyn (especially) and Tillis have said that their hesitancy in approving Blanche had a lot to do with the weaponization fund. Blanche’s approval appeared contingent on a theatrical claim that the fund wasn’t going forward. To have Trump then come out and say “hell yes it’s going forward, even over the objections of Cornyn & Tillis” seems unlikely to appease those two Senators.

Tillis, for one, was not impressed:

That’s Tillis tweeting:

Despite comments as late as yesterday that the fund is dead, President Trump clearly intends to resurrect the payout pot for punks either by inappropriately establishing another bogus fund or pushing Congress to vote for a bill that the majority of Republicans in the Senate would be against.

It’s unfortunate that Todd Blanche, who I consider qualified for the job, will not be confirmed because of this reversal. Hopefully, we can resolve this by Tuesday.

Advertisement

It’s been quite clear that Trump has no intention of ever letting go of this slush fund, but it’s classic Trump that he couldn’t even keep his mouth shut long enough to pretend the fund wasn’t going forward and to get Blanche approved.

Which then brings us to the last bit that came out late Sunday: Blanche tweeting out that the DOJ had finally decided to “rescind” the weaponization fund in writing. Something he had refused to do over the past few weeks.

There are two separate letters there. The first “rescinds” the fund, while making an obnoxious snarky comment about “frivolous lawsuits” filed to challenge it:

A. The Attomey General’s May 18, 2026 Order establishing the Anti-Weaponization Fund (“Fund”) is rescinded and shall have no force or effect.

B. For the avoidance of doubt, nothing in this Order is intended to contradict or otherwise be contrary to prior representations by the Department of Justice that the Fund is not operative. No Members were appointed; no funds were transferred; no process for receiving claims was established; no claims were paid. Nevertheless, several frivolous lawsuits have been filed challenging the Fund, and at least one court has declined to dismiss those claims as moot. This Order establishes, beyond any doubt, that there is no Fund…

If the cases were “frivolous” then why would you need to rescind the agreement?

The second, with snide shots at Congress for not believing his statements, says a bit more on the matter:

Department of Justice Rescinds Anti-Weaponization Order and Addresses the May 19 Release

The Department of Justice today announced that the Acting Attorney General rescinded the May 18, 2026 Order that established “The Anti-Weaponization Fund.” Although the Acting Attorney General has repeatedly advised Congress through testimony, including under oath, as well as in written responses, that the Fund is not moving forward, and the Department has repeatedly represented to district courts that the Fund is not moving forward, today’s Order officially rescinds the May 18, 2026 Order.

Further, the May 19, 2026 Order regarding a mutual release of claims applies by its terms only retroactively. Additionally, the terms “Lawfare and/or Weaponization” in that Order were already defined in the Settlement Agreement as the use of government power “to target individuals, groups, and entities for improper and unlawful political, personal, and/or ideological reasons.”

The Acting Attorney General stands by all of his July 15, 2026 Senate Judiciary Committee testimony, including answers to questions asked by Senator Cornyn related to the scope and applicability of the May 19, 2026 Order. A transcript of the Acting Attorney General’s testimony is attached. He incorporates that testimony by reference, and restates that it is DOJ’s interpretation that the Order only has any effect, including on any release of claims, on the named parties in the lawsuit referenced in the Order.

Advertisement

This all seems pretty obviously designed to get Cornyn and Tillis off his back by saying “see?!? I’ve now put it in writing. Now shut up and vote to make me Attorney General.” And, who knows, it might work.

That’s even though nothing in there actually answers Cornyn’s actual question discussed above. The DOJ cannot single-handedly go back on this agreement. It was, we are told (including in the appeal from Trump on Friday), a private agreement between two distinct parties: Donald Trump and the Justice Department. As such “rescinding” it from just the DOJ side is meaningless since, as Blanche himself admitted to Cornyn, Trump can still try to claim in court that the settlement is valid.

And given Trump’s public statements about why he needs the fund, as well as his lawyers’ statement in the appeal that it tramples his rights to order him not to claim the settlement is valid in court… you’d have to be pretty slow not to pick up on what Trump intends here.

Also, notice what didn’t get rescinded. The May 18th order (creating the slush fund) is what he pretends is gone, but the May 19th order, which clears Trump, his family, and their businesses from IRS audits over past business practices remains. Blanche pretends that saying it “applies by its terms only retroactively” is a meaningful limitation, but that was always obviously the case. By all accounts, the Trump family books are a mess, and they likely owe the IRS a ton of money. And the Blanche announcement makes it clear that remains as is.

So, to sum it all up, Blanche has rescinded an order based on a settlement he can’t single-handedly rescind, to appease Senators whose votes he needs, while the President has made it clear he’s going to move forward with his fund no matter what.

The simple reality here is that we have a crooked President with his hand-picked henchman trying to do the bare minimum to appease a compliant GOP-run Senate, while the President himself can’t keep his own mouth shut long enough to even put up the pretend facade that there’s no corruption going on here.

Under any other President, this would be an impeachable offense. Here it was just another weekend under the Trump Presidency.

Filed Under: anti-weaponization fund, corruption, donald trump, john cornyn, kathleen williams, thom tillis, todd blanche, weaponization fund

Rosetta Stone Sapphire combines over 30 years of language-learning expertise with immersive technology designed to help you speak naturally and confidently in real-world situations. Trusted by millions of learners worldwide, Sapphire brings together structured lessons, pronunciation coaching, conversation practice, flashcards, and personalized learning tools in one complete platform. Built around Rosetta Stone’s proven Dynamic Immersion method, Sapphire teaches you through context, imagery, listening, reading, and speaking instead of relying on direct translation or memorization drills. As you progress through bite-sized lessons, the platform gradually develops your vocabulary, grammar, pronunciation, and conversational confidence through practical everyday scenarios. With access to 25 languages, Rosetta Stone Sapphire also includes advanced learning tools like TruAccent speech recognition, immersive Chat Missions, personalized Sapphire Studio activities, and 500+ flashcard sets covering travel, business, hobbies, culture, sports, and more. Whether you’re preparing for travel, career growth, school, or personal enrichment, Sapphire helps you build practical language skills designed for real conversations. A one year subscription is on sale for $127.20.

Note: The Techdirt Deals Store is powered and curated by StackSocial. A portion of all sales from Techdirt Deals helps support Techdirt. The products featured do not reflect endorsements by our editorial team.

John Ternus takes over as CEO of Apple on September 1, and he’s busy getting his executive team together. The latest addition is Laura Legros, retired VP of hardware engineering.

All eyes are on John Ternus as he prepares to replace Tim Cook as CEO of Apple. Very little is known about the enigmatic character beyond his accomplishments at Apple, but analysts are already looking for challenges he may face.

Every hire and leadership position is being scrutinized, and Bloomberg has learned the latest addition to team Ternus. Laura Legros is coming out of retirement to report directly to Ternus in a vice president role, though the name of her new position wasn’t shared.

She was previously a vice president of hardware engineering and worked for Ternus. All that was shared of her position was a fairly ambiguous statement that she would work “cross-functionally with different parts of the company.”

The only way to interpret that for now is to assume that she’ll be helping ensure the separate hardware teams aren’t too siloed in their operations. Johny Srouji is set to be the newly created Chief Hardware Officer, which means hardware engineering and hardware technologies are now under one leader.

While Legros will report directly to Ternus, it seems she’s likely to be structurally under Srouji as some kind of VP in hardware. Basically, even with the new leadership structure, Legros is back as a VP of hardware engineering, even if that wasn’t explicitly stated in the report.

For those keeping track, Legros was one of the executives that retired in 2022 and was included in Bloomberg‘s reporting about an executive exodus at Apple. No, not the current reportedly nightmare scenario, the previous one we were supposed to be concerned about.

Don’t expect John Ternus to suddenly change the face of Apple overnight, if ever. Some minor things might pop up in the first few months, but it’ll be hard to attribute any specific changes to Ternus or plans made under Cook for some time.

Chinese e-commerce and cloud giant Alibaba’s famed Qwen team of AI researchers last night unveiled Qwen3.8-Max, a new flagship 2.4-trillion-parameter mixture-of-experts (MoE) multimodal large language model (LLM) that targets one of the most competitive corners of the frontier AI market: autonomous software engineering and long-horizon enterprise work.

If the company’s published benchmarks hold up under broader independent testing, Qwen3.8-Max doesn’t merely compete with today’s leading proprietary models — it surpasses several of them on some key benchmarks in agentic computing.

Most notably, Qwen reports that Qwen3.8-Max scores 86.1 on the OSWorld-Verified benchmark measuring how well ahead of GPT-5.6 Sol Max (83.2) and Fable 5 (85.0), while also posting the highest reported score on PaperBench and leading or remaining highly competitive across software engineering, research reproduction, multimodal reasoning, and visual web development benchmarks.

The release also signals a potentially significant strategic shift for Alibaba: the company says open weights for Qwen3.8-Max will be released next week, alongside Qwen3.8-27B.

If that happens under a permissive license, it would represent the first time a Max-class Qwen model becomes available for self-hosted deployment—a move that could substantially reshape enterprise adoption.

One important caveat remains, however: Alibaba has not yet disclosed the licensing terms, leaving open the possibility that the release could use a more restrictive custom license, as we saw recently with Chinese rival Moonshot’s open Kimi K3 frontier model, rather than a broadly permissive one such as Apache 2.0.

A different definition of ‘frontier’

Over the past year, the competitive landscape for foundation models has become increasingly specialized.

OpenAI has largely focused its GPT series on general reasoning, multimodal interaction and enterprise productivity.

Anthropic’s Claude series has emphasized coding and dependable long-context reasoning. Google continues to push Gemini toward multimodal productivity and web-native workflows.

Moonshot AI’s Kimi K3 recently entered the conversation by pairing frontier-class performance with an open-weight release.

Qwen3.8-Max attempts to combine many of these strengths into a single model aimed squarely at enterprise automation.

Rather than emphasizing conversational intelligence, Alibaba is positioning the model as an autonomous coworker capable of executing projects that span days rather than minutes.

According to the company, Qwen3.8-Max can autonomously complete software projects lasting more than 10 days, reproduce research papers involving thousands of lines of code, perform iterative chip-design optimization, and continuously revise plans using multimodal feedback loops.

Those demonstrations remain company-produced and have not yet been broadly replicated by independent evaluators. Nevertheless, they illustrate a growing industry trend: frontier models are increasingly competing on their ability to finish entire workflows rather than answer individual prompts.

Benchmarks increasingly reward autonomous execution

The benchmark suite released alongside Qwen3.8-Max reflects this shift.

Instead of focusing solely on traditional reasoning exams or coding puzzles, many of the highlighted evaluations measure long-horizon execution.

On OSWorld-Verified, which evaluates computer-use agents interacting with desktop environments, Qwen3.8-Max posts 86.1, ahead of GPT-5.6 Sol Max’s 83.2, Fable 5’s 85.0, and Gemini 3.1 Pro’s 76.2.

The model also leads:

• PaperBench: 93.0

• TerminalBench 2.1: 86.6

• Vision2Web: 69.0

• LVBench: 81.8

• ERQA: 77.8

Elsewhere, it remains competitive with proprietary leaders while trailing in several categories.

On the professional software engineering benchmark SWE-Pro, for example, OpenAI’s model posts the highest reported score, while Opus 4.8 continues to lead on certain software engineering evaluations and Agents’ Last Exam.

Rather than dominating every benchmark, Qwen appears to offer one of the broadest balanced performance profiles currently available.

That balance may ultimately matter more for enterprise buyers than isolated benchmark wins.

Many organizations increasingly evaluate models based on how reliably they complete heterogeneous workflows—writing code, reading documents, navigating interfaces, generating reports, inspecting images and coordinating multiple subtasks—rather than optimizing for one narrow capability.

Where Qwen3.8-Max appears strongest

Assuming Alibaba’s published results translate into production deployments, several enterprise workloads stand out as particularly well suited for Qwen3.8-Max.

1. Long-running software engineering

Alibaba’s primary demonstration involves autonomous software development extending beyond ten days.

While enterprises should treat these demonstrations as vendor claims until independently reproduced, they align with a growing interest in persistent coding agents that operate continuously rather than interactively.

Organizations experimenting with autonomous engineering teams, CI/CD automation, repository maintenance, regression testing or feature implementation may find Qwen particularly attractive if its agentic performance proves consistent outside laboratory settings.

2. Computer-use agents

The strongest differentiator may be computer use.

OSWorld has rapidly become one of the industry’s most closely watched benchmarks because it measures a model’s ability to interact with operating systems instead of simply generating text.

Models capable of reliably navigating desktop software can automate countless repetitive business processes, including document processing, enterprise software integration, internal operations and legacy workflows where APIs may not exist.

Leading OSWorld could therefore translate into real operational advantages if benchmark performance generalizes to production environments.

3. Research automation

Qwen’s PaperBench leadership suggests strong potential for organizations performing scientific computing, literature review, experiment reproduction and technical analysis.

Research institutions, pharmaceutical companies and industrial R&D teams increasingly use LLMs not only for summarization but also for executing reproducible computational workflows. Models capable of maintaining context across extended sessions become increasingly valuable in these environments.

4. Multimodal industrial workflows

Unlike earlier multimodal systems that primarily analyze uploaded images, Qwen describes vision as an ongoing feedback mechanism integrated into planning and execution.

That architecture could prove particularly useful in manufacturing, logistics, engineering inspection and design review, where visual inputs continuously inform operational decisions rather than serving as isolated prompts.

The economics may prove just as important

Perhaps the biggest competitive pressure comes not from benchmark scores but from pricing through Qwen’s application programming interface (API) on QwenCloud (based in China):

Qwen3.8-Max launches at $2/$6 per million input/output tokens, a mid-priced model but undercutting the top U.S. proprietary offerings to which it is benchmarked against by meaningful percentages, less than 1/3 the combined in/out price of Claude Opus 5 and less than 1/4 the price of GPT-5.6 Sol Max.

Lower inference costs increasingly matter because agentic systems consume dramatically more tokens than conventional chatbots — a reality that likely factored into OpenAI’s decision late last week to cut the API prices of its mid- and lower-end GPT-5.6 lineup of models (Terra and Luna) by 20% and 80%, respectively.

Indeed, as those running these systems can attest, multi-hour autonomous workflows, iterative planning and continuous self-correction can generate millions of tokens during a single task.

For enterprises deploying hundreds or thousands of agents simultaneously, inference costs often become one of the largest operational expenses. Small reductions in per-token pricing therefore compound rapidly.

How it compares with American frontier models

Despite headline benchmark comparisons, Qwen3.8-Max should not necessarily be viewed as a wholesale replacement for leading American models.

Instead, its strengths suggest different deployment strategies.

OpenAI’s GPT family continues to excel as a broadly capable enterprise reasoning platform with mature tooling, ecosystem integration and extensive commercial deployment. Organizations already invested in Microsoft ecosystems or OpenAI’s enterprise offerings may continue to value those operational advantages even if Qwen leads on selected agent benchmarks.

Anthropic’s Claude Opus remains widely regarded as one of the strongest coding assistants, particularly for careful software engineering and long-context reasoning. Some enterprises may still prefer Claude for human-in-the-loop development where reliability and predictable behavior outweigh raw autonomy.

Google Gemini continues to differentiate itself through deep Workspace integration, multimodal capabilities and Google Cloud services, making it attractive for organizations already standardized on Google’s enterprise stack.

Where Qwen appears most compelling is for enterprises prioritizing autonomous execution, extended planning horizons and favorable inference economics without sacrificing frontier-level performance.

The open-weight question remains unanswered

The largest unknown surrounding Qwen3.8-Max has little to do with benchmarks.

Alibaba says open weights are coming next week. However, neither the announcement nor the provided documentation specifies the license that will govern those weights.

That distinction could prove critical.

A permissive license such as Apache 2.0 would significantly broaden enterprise adoption by allowing organizations to self-host, fine-tune and integrate the model into proprietary products with relatively few restrictions.

A custom license—similar to approaches used by several recent frontier releases—could impose limitations on commercial deployment, redistribution, field of use or model modification. Such restrictions would narrow the appeal for enterprises seeking long-term infrastructure investments, regardless of the model’s technical performance.

Moonshot AI’s recent Kimi K3 release illustrates why this distinction matters. While Kimi K3 made its weights openly available to all, its licensing terms included specific terms including a disclosure and a commercial license requirement for those offering it as a “Model as a Service.”

Until Alibaba publishes Qwen3.8-Max’s license, organizations considering self-hosting should treat the open-weight announcement as promising but incomplete.

An increasingly crowded frontier

Qwen3.8-Max arrives during one of the fastest-moving periods in the history of foundation models.

Within weeks, developers have seen major releases from Moonshot AI, OpenAI, Anthropic and others, each emphasizing different strengths: reasoning, coding, multimodality, autonomous agents or economics.

Alibaba’s contribution is notable because it combines competitive benchmark performance, aggressive pricing, a million-token context window and a stated commitment to releasing weights for its flagship model.

Whether it becomes the preferred platform for enterprise autonomous agents will ultimately depend less on leaderboard positions than on broader independent validation, production reliability and the licensing terms accompanying the forthcoming weight release.

Those factors—not benchmark charts alone—will determine whether Qwen3.8-Max becomes a genuine alternative to the leading American proprietary models or simply another impressive entrant in an increasingly crowded frontier AI race.

from the locking-up-the-internet dept

The Senate Commerce Committee is set to consider S. 737, the SCREEN Act, a sweeping age-verification bill that would require online services to verify users’ ages before they can access any sexually explicit content. If this bill passes, it will force millions of adult internet users to give up their anonymity, privacy, and security before they access lawful speech.  

Unlike many state-age verification laws—which have been harmful in their own right—the SCREEN Act has no requirement that a significant portion of the website consist of sexually explicit content that is harmful to minors. The bill requires nearly any service hosting even a single piece of sexually explicit content to verify the ages of its users. The result is that the bill would apply not only to adult content sites like PornHub or OnlyFans, but also streaming services like Netflix, and social media platforms like Reddit, Discord, or Bluesky, if they host any adult content.

The SCREEN Act does not merely require users to attest they are adults. It specifically states that “requiring a user to confirm that the user is not a minor shall not be sufficient.” In practice, that means platforms would have to verify users’ ages using methods tied to their real identities. Providing proof of age online is dramatically different, and far more invasive, than showing your ID at the door to a bartender or bouncer. In the physical world, the bouncer at the door looks at your ID card, confirms you’re old enough, and gives it back to you. Under the SCREEN Act, the “bouncer” will be a digital age-verification service that captures your personal information and saves it to a database for an unspecified amount of time. 

The consequences of the bill won’t be limited to minors. If websites and apps are expected to reliably identify teenagers, adults will be asked to prove they are adults. 

Even worse, the SCREEN Act is a privacy and data security nightmare. One provision of the bill requires services to take reasonable steps to protect the data collected and to not maintain for longer than is necessary. But these are terribly weak protections that impose no meaningful collection, use, or retention limits on services collecting people’s private information. 

In other words, the third parties tasked with verifying a user’s age on a platform could sweep up a lot of personal info they don’t actually need and then could use that information for any number of purposes, so long as they deem their actions reasonable. Companies would then be allowed to keep the information users have been compelled to turn over for as long as possible, raising security and privacy issues along the way.

The SCREEN Act Attacks Your Right To Use VPNs

The SCREEN Act also targets virtual private network (VPN) users and providers. The bill requires covered websites to verify users’ ages based on their IP addresses unless the service can determine that the user is outside the United States, and specifically requires age verification on traffic coming from known VPN addresses. In practice, this discourages the use of VPNs and proxy servers, which millions of people rely on for legitimate purposes such as protecting personal privacy, securing public Wi-Fi connections, safeguarding journalists and activists, and preventing data tracking. 

VPNs mask your real location by routing your internet traffic through a server somewhere else. When you visit a website through a VPN, that website only sees the VPN server’s IP address, not your actual location. It’s like sending a letter through a P.O. box so the recipient doesn’t know where you really live. VPNs are a privacy and security tool used by millions of internet users every day, and their use should not be treated as suspect. It is particularly galling that the SCREEN Act forces users who intentionally take steps to protect their privacy to identify themselves.

The SCREEN Act creates onerous age-verification rules that will block adults from accessing lawful speech, curtail their ability to be anonymous, and jeopardize the data security and privacy of all internet users. 

Republished from the EFF’s Deeplinks blog.

Filed Under: age verification, data security, free speech, privacy, screen act, vpns