Connect with us
DAPA Banner
DAPA Coin
DAPA
COIN PAYMENT ASSET
PRIVACY · BLOCKDAG · HOMOMORPHIC ENCRYPTION · RUST
ElGamal Encrypted MINE DAPA
🚫 GENESIS SOLD OUT
DAPAPAY COMING

Tech

How Shopify built an AI stack that doesn’t care which models survive

Published

on

Shopify built an LLM proxy that gives every engineer access to multiple AI providers — with automatic failover when any one of them goes down, changes, or disappears. When Claude Fable 5 shut down, Shopify’s engineers didn’t go into panic mode. The proxy shifted them to Claude Opus or GPT 5.5 automatically, without interrupting their workflows.

“Fable looks amazing; we used it of course,” Farhan Thawar, Shopify’s head of engineering, says in a new VentureBeat Beyond the Pilot podcast. “When a model comes and then it goes, or it could be as innocuous as an update, the proxy allows us to spray across the different providers,” Thawar says.

Shopify buys tokens in bulk and all users connect to models through its proxy, Thawar says. This gives his team access to reporting and failover; when there’s an availability issue with one provider, users can be “automatically, seamlessly” transferred to another.

Advertisement

Enterprises can learn from this example and consider how a disruption might affect their business, Thawar says. At the very least, they should establish a solid backup plan. It’s important to have a system that allows for movement across models so enterprises are not “super tied” to a specific provider.

Distillation is another important strategy.

With distillation, a student model learns from a teacher model and typically becomes specialized in a narrower task. These small language models (SLMs) can be more beneficial than generalized, off-the-shelf models in some circumstances. For instance, Shopify’s flagship AI assistant, Sidekick, which performs numerous specialized subtasks for merchants so they can “remove toil” from their day-to-day.

Using smaller distilled models can be faster and cheaper than more generalized models, Thawar says. In some cases they have proven to be 2x cheaper and faster; in more extreme cases 30x cheaper and faster, he says.

Advertisement

But “it isn’t just about cost and latency, which are big; it’s about accuracy,” Thawar says.

Engineers feed the UDP their teacher model, training data, evals, and a target model — say, Opus 4.8 distilling down to Qwen 3.5. The pipeline runs for about a day, then returns an evaluation showing what the fine-tuned model actually achieved on speed, cost, and accuracy for that subtask. If the tradeoff looks good, the engineer deploys it — no approval process required. Shopify’s internal platform, Tangle, lets anyone visualize the pipeline as it runs.

Thawar says his “dream” is to eventually not give the distillation pipeline a target model at all. Instead, users could provide the teacher model with data and evals and the directive: ‘Based on your learnings over time, I want you to look at a different class of model, different sizes, different types, and you tell me what the right distillation target is.’

“Maybe we’ll get surprised. Maybe it’ll be such a small model it could run on a phone,” Thawar says. “Other times, maybe it comes back and says, ‘There isn’t a way to distill this down to anything better than what we have at the frontier.’”

Advertisement

Moving away from “AI reflexivity” to “AI leverage”

Shopify users can apply whatever harness they want: Claude Code, Codex, Cursor, GitHub Copilot for VS Code. “We expose everyone to the different harnesses so they can get a feel for what may or may not work in their workflow.”

But the company also implemented a usage dashboard; this allows Thawar’s team to ask interesting questions around not just token spend, but: Who’s using the most expensive tokens? Who’s spending more time on reasoning? What types of models are being used, and what disciplines and levels?

Regarding the “tokenmaxxing” question, Shopify does have “circuit breakers” in place. If a user has a model running for a long time (say, 10 hours) and it’s consuming a lot of tokens, they will get pinged, “Did you mean to spend this?”

As Thawar explains, sometimes the reply is “Oh, absolutely.” Other times it’s: ‘Whoa, I didn’t know that was running in the background. I totally forgot about it. I’d rather stop it now.’

Advertisement

The ultimate goal, as Thawar describes it, is to move from “AI reflexivity” to “AI leverage,” and get people to really think deeply about where they can benefit most from AI in their workflows.

Listen to the full podcast to hear more about:

  • Shopify’s philosophy of building infrastructure before features. As Thawar puts it: “We’ve always built more infra. We will continue to always build more infra.”

  • How Shopify’s internal AI agent, River, creates a “substrate of information” across the company.

  • How Thawar’s OpenClaw agent figured out he was traveling from his calendar — and what that moment told him about where agents are actually headed.

You can also listen and subscribe to Beyond the Pilot on Spotify, Apple or wherever you get your podcasts.

Source link

Advertisement
Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Tech

A Look At A Gaggle Of Transputer Boards

Published

on

A long time before Beowulf clusters wired up with commodity Ethernet hardware became a hobbyist thing and a running joke, the transputer took a swing at a very similar architecture. This used stand-alone computers that were networked together with other transputer systems, to achieve task-level parallelism. For some people like [Lance Harvie] this is the kind of hardware that he used during his university years for a project, with him not only still having that hardware, but also recently adding to this collection with a recent eBay purchase.

The transputer story is a fascinating one, forming a major part of the UK’s semiconductor industry during the 1980s, creating a strong legacy as the computer industry awkwardly tried to figure out what types of parallelism to target. Whereas the industry largely moved to instruction-level (superscalar) parallelism alongside tightly coupled task-level parallelism along with multiple CPU cores on a single die, one could consider today’s supercomputer clusters to be one example of the transputer legacy.

Close-up of the T424-based 4-processor board. (Credit: Lance Harvie, YouTube)

[Lance]’s university-era board features the T400, which he shows off while recalling programming it in the Occam language. He’s currently looking for an ISA-to-USB adapter to be able to use it again with a modern PC. While searching around, he came across an EBay listing for a four-processor board, containing four T425s. These are significantly more powerful and also can use external memory, unlike the T400.

This four-CPU board omits the external serial links, as it’s meant to be used in e.g. a scientific instrument as a stand-alone 4-unit transputer system, with all of the available four serial links per processor connected on the PCB. Even more interesting is that the processors on this board were manufactured in 1999 by ST, which was many years after transputers stopped being developed.

Advertisement

As [Lance] explains, this was due to the UK government pulling the plug on the transputer project, with the IP subsequently ending up at ST who kept producing the chips until 1999 at its Philippines plant.

In time, [Lance] hopes to power up all these boards and use them again in combination with a modern-day Linux-based computer. We’re definitely looking forward to seeing that happen.

Although you can definitely use any random MCU these days as your very own transputer module or link chip, with e.g. SPI making for an attractive alternative for the high-speed serial links, there’s always something to be said for using real, original hardware.

Advertisement

Source link

Continue Reading

Tech

Short-form science: University of Washington researchers launch PaperTok to combat AI slop

Published

on

Screenshots from PaperTok-generated videos. (PaperTok.com Images)

Researchers have a new weapon against the scientifically inaccurate AI slop muddying public understanding of complex topics. A University of Washington team is helping scientists tell their own stories with a free tool that converts dense, jargon-heavy publications into short, accessible videos.

“There’s a lot of science communication happening in short form — primarily on TikTok, but also we’re seeing YouTube Shorts and Instagram Reels — these tidbits of science findings,” said Meziah Ruby Cristobal, a UW doctoral student in human centered design and engineering.

Cristobal and her colleagues built PaperTok hoping to use AI for good — to fight the technology’s irresponsible use elsewhere by non-scientists who misrepresent research.

The tool is simple. A researcher uploads a paper into PaperTok, which analyzes it to find attention-grabbing hooks and the most relevant takeaways for a general audience. The tool generates a script with an opening scene and narrative arc, producing a 45-second AI-narrated video. It closes with a reference to the paper, including the researchers’ names and the journal, to establish credibility.

Other tools can turn PDFs into videos, but Cristobal said PaperTok was intentionally designed to keep humans in the loop. It uses a multi-step process that requires approval at each phase, giving users the ability to edit the output down to individual words.

Advertisement

Cristobal presented research on PaperTok this spring in Barcelona at the Association for Computing Machinery’s Conference on Human Factors in Computing Systems. She co-led the study with fellow doctoral student Donghoon Shin; the senior author is UW professor Gary Hsieh.

Lead contributors on the University of Washington’s PaperTok study, from left: Gary Hsieh, UW professor in human centered design and engineering; and UW doctoral students Meziah Ruby Cristobal and Donghoon Shin. Cristobal presented the findings at the Association for Computing Machinery’s Conference on Human Factors in Computing Systems in Barcelona. (Photo courtesy of Cristobal)

A team of eight built PaperTok last summer, starting with interviews with science communicators and researchers before developing the tool and gathering user feedback.

“A lot of the researchers actually found huge value in seeing how the AI tries to visualize what they believe to be very abstract concepts,” Cristobal said. For many, it served as a brainstorming tool that highlighted new ways to communicate their findings.

There was critical feedback as well. Some users said the videos felt “too AI-ish,” pointing to issues like nonsense text. The UW team is continuing to refine PaperTok, including plans to let researchers incorporate charts and graphics from their papers into the videos.

PaperTok was built to translate research papers on human-computer interaction but has been tested on topics including physics, and it held up well. The team wants to expand its reach across research disciplines to create videos for social sciences and hard sciences alike.

Advertisement

The tool is free to use, but because video generation is computationally expensive, the company asks researchers to use a Gemini key so the cost is charged to their Google account.

Source link

Continue Reading

Tech

Samsung’s new budget phone Galaxy A27 5G costs $50 more, yet downgrades two key features

Published

on

Samsung just unveiled its newest budget phone, Galaxy A27 5G, starting at $349, which is $50 pricier than last year’s Galaxy A26, yet it actually takes a step back in a few areas that matter for everyday use.

What you actually get for the higher price

Samsung swapped the A26’s Exynos chip for a Snapdragon 6 Gen 3, which delivers a 10 to 20% speed boost, plus a GPU upgrade for smoother gaming and graphics work. The display still rocks a 6.7-inch Super AMOLED panel with a 120Hz refresh rate. It now has a punch hole camera cutout instead of last year’s teardrop notch, with slimmer bezels around the screen.

You get three memory configurations to choose from too, 6GB plus 128GB, 8GB plus 128GB, and 8GB plus 256GB. AI features get a small boost too, including multi-object recognition in Circle to Search, sharper Object Eraser results, and real-time translation in the Voice Recorder app across 22 languages. Samsung is sticking with six years of OS upgrades and security updates, just like before. However, most of this feels like a minor refresh rather than a real leap forward.

Galaxy A27 5G vs A26: what’s downgraded?

The biggest letdown is IP ratings. Samsung Galaxy A26 was the first in its tier to score an IP67 rating, meaning it could survive a brief dunk, and Samsung loved bragging about that flagship-grade feature trickling down to a budget phone. Sadly, the A27 drops to IP64, which only shrugs off splashes and dust, not a dip in water.

Meanwhile, the cameras took a hit too. The ultrawide lens falls from 8MP to 5MP, and the selfie camera drops from 13MP to 12MP, though the 50MP main camera with OIS stays the same. Additionally, the phone is even slightly thicker now, going from 7.7mm to 7.8mm, which is a negligible difference, but a regression nonetheless.

Advertisement

If you are still looking forward to picking one up, the Galaxy A27 5G lands July 3 internationally and July 14 in the US. And if the price hike has you eyeing other options, here are four terrific Galaxy A37 alternatives worth checking out too.

Source link

Advertisement
Continue Reading

Tech

Notion Mail shuts down amid agent takeover

Published

on

Productivity company Notion is shutting down its email product, Notion Mail, on September 22. The company said it is discontinuing its email inbox in favor of its AI agent offering. It noted that users were increasingly handing over reins of their email to the agents, and not opening their inbox at all.

“As Notion agents have gotten more capable, we’ve seen more users hand off email workflows to them. Today, more than half of Notion Mail users manage emails without ever opening their inbox. So, we’re going all in on using agents to run your inbox,” the company said in a post on X.

Notion Mail is connected with Gmail, meaning all emails in the inbox will stay intact. However, users will need to export drafts and scheduled emails if they want to keep them. The company said that users can export snippets and auto-label instructions and use them elsewhere and emphasized that Notion’s email-based agents will keep working post-Notion Mail shutdown.

Notion announced its email product in preview mode in 2024 after it acquired security-centric productivity startup Skiff. The company aimed to integrate email with Notion AI with features like auto-labeling, filtering, and handling scheduling for users. The company made the product available to users in April 2025 to better compete with the likes of Superhuman and Fyxer. Newer startups like AgentMail are in step with Notion’s thesis and are trying to build an email service specifically for agents.

Advertisement

Source link

Continue Reading

Tech

CSS On The ESP32 | Hackaday

Published

on

There are lots of graphics libraries available for the ESP32, and lots of ways to program one to boot. Even still, most of us wouldn’t immediately think to CSS when it comes to embedded products — yet that’s now a thing on the Espressif platform, apparently.

The Gea stack allows one to compose CSS and TypeScript code that is then turned into generated C++ code that compiles to native firmware. The team behind Gea have demoed this ability by running a 3D cube animation on an ESP32 at up to 60 FPS. This isn’t some ugly, low-res wireframe demo, either. It’s a full-color animation running on a 410×502 AMOLED screen. It’s very fluid, and can even handle transparency on the cube faces (albeit with a performance penalty).

It’s worth noting that this isn’t a full browser engine. As you might expect, some concessions had to be made to get it running on the ESP32. Namely, it doesn’t handle “:hover” states because it’s designed for touchscreen use, fonts are rasterized, and the UI tree is limited to just 512 nodes. Regardless, it shows that using CSS and TypeScript to develop for the ESP32 is entirely possible without some crazy loss of performance. If you want to build easy interfaces on an ESP32 while leaning on web dev experience, this could be very useful indeed.

Advertisement

There are lots of fun ways to write code for the ESP32; you can even try MicroPython if you like.

Advertisement

Source link

Continue Reading

Tech

As Xbox gets pricier, Microsoft launches Buy Now, Pay Later scheme for consoles

Published

on

Earlier today, Microsoft raised the price of its Xbox consoles by up to $150 in the US. Following the price hike, the asking price for the Xbox Series X 2TB edition has climbed all the way up to $800. The 1TB model now costs $650, while the Xbox Series S with 512 GB storage will now cost $400 in the US market. 

What’s the game plan?

To retain gaming enthusiasts, Microsoft has introduced a new installment-based payment scheme for its consoles that is now live through its official storefronts. The Buy Now Pay Later system is currently available for the Xbox Series S and the Xbox Series X consoles — for both new and refurbished units. 

Now that the price of Xbox has gone up by up to $150, Microsoft is introducing Buy Now, Pay Later option.
“We’ve made it easier for players to use Buy Now, Pay Later options on eligible XBOX hardware purchases through Microsoft Stores, making it possible to break up your payment… pic.twitter.com/6PgLHsEumw

— Digital Trends (@DigitalTrends) June 25, 2026

According to Microsoft’s website, the Buy Now, Pay Later system works in collaboration with PayPal and allows you to pay the full cost of a device in four installments (paid bi-weekly) with zero interest, or you can opt for monthly installments and pay them across installments.

Advertisement

If you opt to pay monthly for an Xbox console, you can break the installments across 24 months. There are no late fees or sign-up charges involved. “Flexible payment options help you pay how you want, at Microsoft Store,” says the company on its online storefront.

The caveat, and alternative

But in order to avail the buy now, pay later benefit on Microsoft Store, which is essentially an extension of the PayPal Pay Later system, your purchase must first be approved before you can decide the duration over which you want to pay the full cost of the Xbox console. 

The situation with memory pricing is pretty grim. Earlier today, Apple raised Mac and iPad prices, after claiming the situation is unsustainable now. #Xbox is running into a similar wall. This is what the company has to say:
“Unfortunately, console storage and memory prices… https://t.co/NERFunpNlv

— Digital Trends (@DigitalTrends) June 25, 2026

In case your purchase is not approved by PayPal or you don’t have a PayPal account, you can head over to Amazon and purchase an eligible Xbox console via a financing scheme with zero interest, spread across 12 months. 

Advertisement

Microsoft won’t be the only player reeling under the stress of rising memory and storage chip prices, ultimately forcing it to raise the hardware prices. Earlier today, Apple also raised the sticker price of Mac and iPad hardware significantly. Elsewhere, the cost of PCs and laptops has gone up, and even Valve had to price its Steam machine at over $1,000 in the U.S. market.

Source link

Advertisement
Continue Reading

Tech

a16z-backed Base Power is offering cheaper electricity to the power grid that needs it most

Published

on

Energy storage startup Base Power began selling its massive home battery systems to residents of Illinois yesterday, Canary Media reported. Crucially, it’ll be the startup’s first foray into the grid territory operated by PJM Interconnection, the largest U.S. grid operator by territory, and one that has particularly struggled to cope with an onslaught of new data centers.

Beyond Illinois, PJM’s territory includes Northern Virginia, one of the densest data center regions on the planet. That density, coupled with a paucity of new generating sources, has caused wholesale electricity prices in PJM to nearly double over the past year. The power crunch has gotten so bad that AEP, one of the region’s largest utilities, has threatened to leave the market.

Base Power launched two years ago in Texas to build a virtual power plant centered around residential batteries. Base’s batteries, starting at 25 kilowatt-hours, are bigger than many of its competitor’s, and rather than sell the batteries, it requires customers to buy electricity from it. In Illinois, its rates are 25% below utility ComEd’s.

The startup’s timing has also been impeccable. Base is currently operating more than 500 megawatt-hours of battery storage in Texas, charging when electricity prices are cheap and dispatching them when the grid needs it most. 

Advertisement

Its entry into the PJM grid comes at a time when the operator has come under scrutiny for bungling its handling of rising electricity demand. PJM had paused applications for new generating sources starting in 2022, only reopening the queue in April. Unlike Base, its timing couldn’t have been worse — electricity demand has skyrocketed in the last four years.

Base’s rollout has gathered pace since October, when it announced a $1 billion round led by Addition. That round followed closely on the heels of a $200 million round which Andreessen Horowitz, Lightspeed Venture partners, and Valor Equity Partners led in April 2025.

Historically, PJM has been slow to adopt new technologies like distributed energy storage, but Base’s residential focus helps it do an end-run around the sclerotic grid operator.

“We are deploying capacity behind the meter at the residential home, where an interconnection already exists, so we don’t wait in the interconnection queue,” Zach Dell, Base Power’s founder and CEO, told Canary Media.

Advertisement

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

Source link

Continue Reading

Tech

iPads just got a price hike, and this is your last chance to save a fortune on them with Prime Day deals

Published

on

Apple has raised prices across Macs and iPads as the ongoing memory and storage crunch continues to push up component costs. Several iPad models are now $100 to $200 more expensive than before. This makes the ongoing Amazon Prime Day deals one of the last easy ways to buy an iPad at a more reasonable price before higher official pricing becomes the new normal. We have listed all the ongoing iPad offers below, so check out these deals before the Prime Day sale ends on June 26.

iPad (A16, 128GB) jumped from $349 to $449, but Prime Day has it for $299

Everyday iPad pick

Pros

  • Clean design and premium build
  • Sharp display for the price
  • Plenty of A16 silicon firepower
  • 128GB storage for the same price
  • Battery life doesn’t disappoint

Cons

  • Accessories are a little too priceyr
  • No Apple Intelligence or Stage Managerr
  • Stylus support is outdated
  • Non-laminated display, againr
  • Slow charging

The iPad (A16, 128GB) used to cost $349, but Apple’s new price is $449 after the latest $100 hike. Amazon’s Prime Day deal brings it down to $299, which is $150 less than Apple’s new price. This is the iPad I’d recommend for most people who want a tablet for streaming, browsing, reading, video calls, casual gaming, and everyday use without paying Air or Pro prices.

iPad mini (128GB) jumped from $499 to $599, but Prime Day has it for $499.99

Small travel iPad

Pros

  • Pocketable size and gleefully lightr
  • Better aspect ratio than other iPadsr
  • Sharp and colorful displayr
  • A17 Pro silicon is plenty powerful
  • Battery life doesn’t disappointr

Cons

  • Screen should be brighter
  • Apple Intelligence is still half-bakedr

The iPad mini (128GB) has also been hit by a $100 price hike, moving from $499 to $599. During Prime Day, Amazon has it for $499.99, which keeps it roughly at its old Apple price and about $99 below the new one. This is the better pick if you want a compact iPad for reading, travel, note-taking, gaming, or one-handed use.

The M4 iPad Air jumped by $150, but these Prime Day deals make the 11-inch and 13-inch models easier to buy

Everyday productivity pick

Pros

Advertisement
  • Clean designr
  • Sharp screen for creativity and productivityr
  • M3 power boost at no added costr
  • More affordable new Magic Keyboardr
  • Good battery lifer

Cons

  • Expensive storage and connectivity upgradesr
  • Display needs better smudge-resistancer
  • iPadOS has its quirksr
  • Accessories are quite pricey

The iPad Air lineup has become $150 more expensive after Apple’s price hike. The 11-inch iPad Air (128GB) has moved from $599 to $749, but Amazon’s Prime Day deal brings it down to $519, which is $230 less than the new Apple price.

The larger 13-inch iPad Air (128GB) has gone from $799 to $949, while Amazon has it for $699.99, saving you about $249 compared with the new price. The 11-inch model is the better pick for students, work, and everyday productivity, while the 13-inch version makes more sense if you want a bigger screen for multitasking, drawing, watching videos, or using the iPad more like a laptop replacement.

iPad Pro jumped by $200, but these Prime Day deals cut the 11-inch and 13-inch M5 models back down

Creator-focused iPad

Pros

  • Stunning Ultra Retina XDR OLED display
  • Thin and lightweight design
  • Excellent tablet for drawing, editing, streaming, and productivity
  • Blazing-fast M5 performance

Cons

  • Very expensive
  • Overkill for casual users
  • Not a huge upgrade if you already have the M4 iPad Pro

The iPad Pro models have seen the biggest jump, with both M5 models now $200 more expensive than before. The 11-inch iPad Pro (M5, Wi-Fi, 256GB) has gone from $999 to $1,199, but Amazon’s Prime Day price is $899, saving you $300 versus Apple’s new price.

The 13-inch iPad Pro (M5, 256GB) has climbed from $1,299 to $1,499, while Amazon has it for $1,199, also saving you $300. The 11-inch iPad Pro is the better choice if you want Pro-level power in a more portable size, while the 13-inch model is the one to get if you want Apple’s largest and most powerful iPad for creative work, multitasking, or replacing a laptop-style setup.

Disclaimer: These iPad deals are subject to change, and availability may be limited due to high demand.

Advertisement

Source link

Continue Reading

Tech

Judge Says Florida’s Social Media Law Is “Literally Impossible” To Obey. Thanks To The Supreme Court, It Gets A Trial Anyway.

Published

on

from the the-moody-reality dept

Remember a few years ago when both Texas and Florida passed laws trying to tell social media companies they couldn’t moderate political content? Those cases eventually made their way to the Supreme Court, where the Court (as it’s been known to do) kinda punted: sending the cases back to the lower courts on technical legal grounds, claiming that the plaintiffs, NetChoice and CCIA, had mistakenly filed the cases as “facial” challenges, rather than “as applied.” It’s not worth going into the legal weeds again about the difference here, especially since the ruling had tons of important and useful language making it clear that content moderation is protected by the First Amendment in the case dubbed “Moody” after Florida’s former Attorney General.

But, the two cases have continued to bounce around the courts over the past few years, and the district court in Florida has rejected both the plaintiffs and the state’s motions for summary judgment, but in doing so has again made some great arguments about how content moderation gets First Amendment protections. The case is still before Judge Robert Hinkle, who made the original ruling finding the law unconstitutional five years ago.

Now, after discovery, Hinkle is reviewing the amended complaint that tries to deal with Moody’s limits on “facial” challenges. He starts out by reinforcing that content moderation is obviously protected by the First Amendment:

Collecting third-party speech content into a single speech product is what social-media platforms do. As the Supreme Court said, the collection “is itself expressive, and intrusion into that activity must be specially justified under the First Amendment.” Id. The defendants’ contrary assertion was rejected in NetChoice I (this court’s preliminary-injunction order), and in NetChoice II (the Eleventh Circuit’s opinion affirming that order in relevant part), and in Moody (the Supreme Court’s opinion agreeing with the Eleventh Circuit in relevant part).

Florida tried to argue that because recommendation engines try to keep users on the platform, and that decisions are made by “algorithms” that somehow changes the equation. The court points out that this is technically illiterate, because humans still set the editorial guidelines:

Advertisement

The defendants say platforms decide which content to show any given user primarily based on the user’s viewing habits, showing the user the content most likely to keep the user on the platform longer. Perhaps so. See Moody, 603 U.S. at 735 (“The selection and ranking is most often based on a user’s expressed interests and past activities.”). And the defendants say this decision is made by algorithms, devoid of human involvement. Not so. Humans adopt the standards and guidelines, establish algorithms that incorporate them, and keep a great deal of content off the platforms on this basis, even though, as the defendants emphasize, the remaining content is organized to a substantial extent by algorithms based on a user’s viewing habits. This record establishes without genuine dispute that the six platforms specifically addressed in the plaintiffs’ motion have standards or guidelines that have a significant role in selection and organization of content. “And because that is true, they receive First Amendment protection.” Moody, 603 U.S. at 740.

Moody reiterated this point in discussing Texas’s similar legislation. The Court said the parties treated Facebook’s News Feed and YouTube’s homepage as the heartland applications of the Texas law—much as those and other platforms’ similar features are the heartland applications of the Florida law. See id. at 744. The Court said that at least on the record then before the Court, “the editorial judgments influencing the content of those feeds” were “protected expressive activity” that Texas could not “interfere with . . . simply because it would prefer a different mix of messages.” Id. (emphasis added). The Court said “influencing,” not “fully determining.” The record now before this court makes clear that editorial judgments of the six platforms addressed in the plaintiffs’ motion at least influence the content of their feeds. The First Amendment applies.

While Justice Barrett made some technically questionable statements in a concurrence about whether AI-driven algorithms might change the equation, Judge Hinkle says that even if she were right, it wouldn’t matter here:

But the defendants are plainly incorrect when they assert, in substance if not explicitly, that the First Amendment does not apply when there is mixed curation—some driven by human editorial discretion and some by algorithms or artificial intelligence. Responding to Justice Barrett’s concurrence, the Court said this case does not deal with “feeds whose algorithms respond solely to how users act online—giving them the content they appear to want, without any regard to independent content standards.” Moody, 603 U.S. at 736 n.5 (emphasis added). The Court continued, “Like them or loathe them, the Community Standards and Community Guidelines make a wealth of user-agnostic judgments about what kinds of speech, including what viewpoints, are not worthy of promotion. And those judgments show up in Facebook’s and YouTube’s main feeds.” Id. Justice Barrett joined that footnote.

And then the key point: the First Amendment protects content moderation. Full stop.

The unmistakable upshot is this: the First Amendment applies to mixed curation. The defendants’ contrary assertion is inconsistent with Moody, the many precedents discussed in Moody, and any coherent view of the First Amendment. This does not mean platforms are not subject to government regulation, but it does mean regulation must pass appropriate First Amendment scrutiny.

The judge also finds that there are constitutional problems with how vague the law is in some areas. And, in others, finds that the law would be impossible to comply with. In discussing the law’s prohibition on “post-prioritization or shadow banning algorithms” for any posts “by or about” a candidate for office, the court finds the provision baffling — saying its plain meaning makes no sense, and that Florida’s defense of it makes even less sense:

Advertisement

But the provision prohibits a platform from using “post-prioritization or shadow banning algorithms” for content by or about a user known to be a candidate. “‘Post-prioritization’ means action by a social media platform to place, feature, or prioritize certain content or material ahead of, below, or in a more or less prominent position than others in a newsfeed, a feed, a view, or in search results.” Fla. Stat. § 501.2041(1)(e) (emphasis added). Unless a platform shuts down completely, compliance with this provision is literally impossible; posts can only be ahead of or below other posts, and posts can only be in a more or less prominent position than other content.

The defendants say, though, that the provision does not mean what it says— that it requires posts by or about candidates to be placed in chronological order. Perhaps Florida courts will rewrite the provision in this way, but they have not done so to this point. One doubts the Florida legislature really intended to require all candidate posts to go to the top, allowing candidates and their supporters to flood every user’s feed, rendering platforms useless, or nearly so. And if that is not what the provision means, one is at a loss to divine any plausible meaning.

Thus, the court says these provisions are likely unconstitutionally vague.

Still, NetChoice/CCIA don’t win their own summary judgment motion, in part because the court says that their amended complaint is still a “quasi-facial challenge” which runs into the same issues the original challenge faced at the Supreme Court, and because of that the court holds off on granting summary judgment, meaning the case continues to move forward to trial, even as the judge makes it pretty clear this law is a complete constitutional mess.

So this is about as good a ruling as NetChoice and CCIA could realistically hope for, given the procedural mess the Supreme Court handed down in Moody. Hinkle has made it abundantly clear that he thinks Florida’s law is a vague, unconstitutional disaster that can’t survive contact with the First Amendment. And yet, because the Supreme Court decided that “facial vs. as-applied” was the hill to die on, he can’t just say so and end it. Instead, a law that everyone — including the judge — can see is unconstitutional gets to march all the way to trial.

Advertisement

That’s the real legacy of Moody’s procedural punt: it didn’t save these laws, but it did make killing them slower, costlier, and more painful than it has any right to be.

Filed Under: 1st amendment, as applied challenge, content moderation, facial challenge, florida, free speech, james uthmeier, moody, moody v. netchoice, robert hinkle, sb 7072, social media

Companies: ccia, netchoice

Advertisement

Source link

Continue Reading

Tech

I can’t believe how cheap the iPad Air M3 is for Prime Day

Published

on

It’s not often we can say an iPad is affordable, but this Prime Day deal makes the iPad Air more appealing to budget-conscious shoppers.

The 13-inch iPad Air M3 is now down to just £650 in this rare price cut. That’s nearly a massive £200 off its usual price, and marks the cheapest the tablet has ever been on Amazon. 

iPad Air M3 on table anglediPad Air M3 on table angled

The iPad Air M3 is seeing nearly £200 off in this rare Prime Day deal

We can’t believe how cheap the iPad Air M3 is for Prime Day. At just £650, not only does it boast a nearly £200 price cut, but this is the cheapest we’ve ever seen the Apple tablet reach on Amazon. Act fast to avoid disappointment.

Advertisement

View Deal

Just note that this model isn’t the latest iteration, as it has since been succeeded by the iPad Air M4. Even so, the iPad Air M3 remains a seriously capable tablet and a brilliant alternative for those who don’t necessarily care about owning the latest model.

Advertisement

Powering the iPad Air is Apple’s desktop-level M3 chip which is able to handle all tasks with ease. In fact, we were even pleasantly surprised at how well the iPad Air performed when gaming, as it supports hardware-accelerated ray tracing in certain titles. In addition, there’s the Apple Intelligence toolkit at your fingertips which, although shouldn’t be the sole reason you opt for the iPad Air, is a nice addition to have.

Otherwise, the iPad Air M3 runs on iPadOS 26 which enables the tablet to work a lot more like a Mac, as you can multitask across apps and move them around much easier than before. If you’re looking for a laptop alternative, then the iPad Air M3 is a hard option to beat.

Advertisement

Speaking of which, you can turn the iPad Air into a makeshift laptop by purchasing the Magic Keyboard and Apple Pencil separately. How useful either accessory really is will depend on your personal usage and preference.

Advertisement

Overall, the iPad Air M3 earned a 4.5-star rating with Editor Max Parker concluding the tablet can do “just about everything the Pro series can do” and hailed it as “all the iPad you need”.

Now at its lowest ever price, the iPad Air M3 is finally an appealing choice for those seeking a tablet but not wanting to spend a fortune in the process.

Advertisement

Source link

Advertisement
Continue Reading

Trending

Copyright © 2025