The introduction of translation to Apple’s Airpods 3 might be useful or.. might not at all. CC-licensed photo by download.net.pl – mobile on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

---

A selection of 10 links for you. Utilisez-les à bon escient. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

---

California limits on “addictive” social media feeds for children largely upheld • Reuters

Jonathan Stempel and Nate Raymond:

»

A federal appeals court largely upheld a California law on Tuesday making it illegal, absent parental permission, for social media companies to provide children with “addictive feeds” that the state fears could damage their mental health.

The 9th U.S. Circuit Court of Appeals rejected most claims by the technology trade group NetChoice, which said California’s Protecting Our Kids from Social Media Addiction Act was overbroad and vague and violated the First Amendment.

Addictive feeds are algorithms that select personalized media for users based on those users’ online behavior.
NetChoice, whose 41 members, opens new tab include Google, Facebook and Instagram parent Meta Platforms, Netflix and Elon Musk’s X, said the law signed by Governor Gavin Newsom last September unconstitutionally limited members’ ability to speak to children through the algorithms.

Writing for a three-judge panel, Circuit Judge Ryan Nelson said the issue of which algorithm-based feeds were “expressive” for First Amendment purposes was fact-intensive, and NetChoice did not show the California law’s alleged unconstitutional applications predominated.

Nelson also found NetChoice premature in challenging a requirement that platforms take steps to verify users’ ages, rather than simply limit feeds to users it knows are children, because the requirement doesn’t take effect until 2027.
The court blocked a requirement that accounts’ default settings prevent children from seeking how many likes and other comments their posts receive. It said that requirement was not the least restrictive way to protect children’s mental health.

Paul Taske, co-director of the NetChoice Litigation Center, said the group is “largely disappointed” with the decision.
“California’s law usurps the role of parents and gives the government more power over how legal speech is shared online,” Taske said. NetChoice has filed many lawsuits challenging state-level internet restrictions.

«

This is surely headed for the Supreme Court, which will maybe rule in the tech firms’ favour, though it probably depends on which way the political winds are blowing at the time.
unique link to this extract

---

Record-breaker: Leonard Barden’s chess column celebrates 70 years and a place in history • The Guardian

Sean Ingle:

»

September 1955. The dying embers of one era, the dawning of another. It’s been five months since Sir Winston Churchill retired as prime minister. In another four Elvis Presley will release Heartbreak Hotel, his first worldwide hit. Food rationing is over. Frozen fish fingers, courtesy of Clarence Birdseye, have just arrived.

Change is also in the air at the Manchester Guardian. On 8 September a young chess master from Croydon, Leonard Barden, writes his first column. His subject is a Russian teenager, Boris Spassky, whose games, Barden notes, “all show the controlled aggression characteristic of a great master”.

The writing is lively and accessible. The judgment impeccable. Spassky will go on to become world champion. Meanwhile, Barden is on the foothills of a journey that, 70 years, 15 prime ministers, and nearly 4,000 articles later, is still going strong.

In all that time he has never missed a week – rain or shine, in sickness and in health. And now, officially, he is a record-breaker. Barden recently eclipsed Jim Walsh of the Irish Times, who started his column in July 1955 and finally retired in May this year, to set a Guinness World Record for the longest-running continuous chess column.

Barden, who recently turned 96, is also the longest-serving daily newspaper columnist for his 63-year stint with the Evening Standard, which ended in 2020. Both records will surely never be broken. Yet they are only a small part of an astonishing career.

He was British chess champion in 1954. He played for England in four chess Olympiads. And he was a key figure in the British chess explosion of the 70s and 80s that turned players such as Nigel Short into world title contenders. As the grandmaster Raymond Keene put it a couple of years ago: “Everywhere you looked in British chess, the giant handprint of Len Barden was to be found.”

«

As with so many stories like this, he nearly came a cropper early on when he was given duff advice and the solution was wrong. The chief sub gave him an official final warning.

Anyway, he’s still writing articles. Each one has a chess puzzle – typically just a “move to win” – from real play which is always entertaining.
unique link to this extract

---

Narrative podcasts are disappearing. What happened? • Rolling Stone

Eric Benson:

»

For years, narrative podcasts were the buzziest segment of the industry. They were what many listeners thought of when they heard the word “podcast.” Many of these shows, following the lead of Serial, were true crime, while others, like Slow Burn, took a similar investigative approach to politics and history. But regardless of genre, most narrative shows shared a style — they were told through the perspective of a single reporter-host, who not only guided a listener through a story but invited them along on the reporting journey.

Now, these shows seem increasingly like an artifact of history. Layoffs have hit nearly every organization that makes this kind of audio, from corporate behemoths like Spotify to successful start-ups like Pushkin Industries to the public-radio godfather of it all, This American Life. In early August, Amazon announced it was dismantling Wondery, one of the biggest and most commercially successful studios in the business, which had made its name on shows like Dr. Death and The Shrink Next Door. Amazon had acquired Wondery less than five years ago for $300m. Now, the company was being broken up into parts and a reported 110 staffers were laid off.

“It feels like podcasts speed-ran the development of an industry to the decline of an industry,” the journalist and podcast host Evan Ratliff told me. “It went from people were making absolutely no money, to people were making fortunes big enough where they’d never have to work again, to ‘There’s not a budget for that’ in less than 10 years.” Now, even those reduced budgets have vanished.

The fall of the industry has been so vertiginous that it’s been hard to fully comprehend its decline. But in many ways, this collapse was baked into its spectacular rise, when a flood of dumb money, pollyannaish entrepreneurs, and hungry journalists rushed to build an industry that would soon turn into a house of cards.

…The tech and media giants that had funded the podcast boom had learned that narrative journalism was hard to scale. In 2019, it might have been enough for them to make shows that “we can brag about,” but a few years later, as recession fears grew, losing money on prestige programming no longer seemed like a good idea.

«

Early entrants tend not to be long-term survivors, lesson 945. Much easier to get a few people in a room to talk about any old thing than make a radio documentary. Or actually.. who needs people?
unique link to this extract

---

AI podcast start up plans 5,000 shows, 3,000 episode a week • Hollywood Reporter

Caitlin Huston:

»

Why pay a celebrity podcast host millions when you can create your own using AI? 

Inception Point AI is attempting to do just that, as the company builds a stable of AI talent to host podcasts, and eventually become broader influencers across social media, literature and more. Amid the high costs for producing narrative podcasts and pricy, short-term contracts for popular hosts, the idea here is being able to own, scale and control the talent (unlike those off-the-cuff humans) and produce shows at a minimal cost. 

“We believe that in the near future half the people on the planet will be AI, and we are the company that’s bringing those people to life,” said CEO Jeanine Wright, who was previously chief operating officer of podcasting company Wondery, which has recently had to reorganize under the changing podcast landscape. 

The company is able to produce each episode for $1 or less, depending on length and complexity, and attach programmatic advertising to it. This generally means that if about 20 people listen to that episode, the company made a profit on that episode, without factoring in overhead. 

Inception Point AI already has more than 5,000 shows across its Quiet Please Podcast Network and produces more than 3,000 episodes a week. Collectively, the network has seen 10 million downloads since September 2023. It takes about an hour to create an episode, from coming up with the idea to getting it out in the world. 

The company produces different levels of podcasts. The lowest level involves weather reports for various geographic areas or simple biographies and higher levels involving subject-area podcasts hosted by one of about 50 AI personalities they’ve created, including food expert Claire Delish, gardener and nature expert Nigel Thistledown and Oly Bennet, who covers off-beat sports.

As for how it stacks up against human podcasts? “I think that people who are still referring to all AI-generated content as AI slop are probably lazy luddites. Because there’s a lot of really good stuff out there,” Wright said. 

«

I think by “in the near future half the people on the planet will be AI” he means we humans will be interacting with AI as often as we do humans. Sure, but we will get a lot more value from the human interaction. And while I’m sure there are people who want to listen to AI-generated and -read podcasts, I’m not one of them.
unique link to this extract

---

Judge: Anthropic’s $1.5bn settlement is being shoved “down the throat of authors” • Ars Technica

Ashley Belanger:

»

At a hearing Monday, US District Judge William Alsup blasted a proposed $1.5bn settlement over Anthropic’s rampant piracy of books to train AI.

The proposal comes in a case where Anthropic could have owed more than $1 trillion in damages after Alsup certified a class that included up to 7 million claimants whose works were illegally downloaded by the AI company.

Instead, critics fear Anthropic will get off cheaply, striking a deal with authors suing that covers less than 500,000 works and paying a small fraction of its total valuation (currently $183bn) to get away with the massive theft. Defector noted that the settlement doesn’t even require Anthropic to admit wrongdoing, while the company continues raising billions based on models trained on authors’ works. Most recently, Anthropic raised $13bn in a funding round, making back about 10 times the proposed settlement amount after announcing the deal.

Alsup expressed grave concerns that lawyers rushed the deal, which he said now risks being shoved “down the throat of authors,” Bloomberg Law reported.

In an order, Alsup clarified why he thought the proposed settlement was a chaotic mess. The judge said he was “disappointed that counsel have left important questions to be answered in the future,” seeking approval for the settlement despite the Works List, the Class List, the Claim Form, and the process for notification, allocation, and dispute resolution all remaining unresolved.

Denying preliminary approval of the settlement, Alsup suggested that the agreement is “nowhere close to complete,” forcing Anthropic and authors’ lawyers to “recalibrate” the largest publicly reported copyright class-action settlement ever inked, Bloomberg reported.

Of particular concern, the settlement failed to outline how disbursements would be managed for works with multiple claimants, Alsup noted. Until all these details are ironed out, Alsup intends to withhold approval, the order said.

One big change the judge wants to see is the addition of instructions requiring “anyone with copyright ownership” to opt in, with the consequence that the work won’t be covered if even one rights holder opts out, Bloomberg reported.

«

Making it opt-in will significantly reduce the number of people who claim – so a bigger payout for those who do. More background from the initial story about the $1.5bn offer last week.
unique link to this extract

---

Apple announces AirPods Pro 3 with “world’s best ANC” and heart rate sensing • The Verge

Tom Warren:

»

Apple just announced the AirPods Pro 3, the first major update to the product in three years. Apple is introducing a new heart rate sensor in the AirPods Pro 3, as well as improved active noise cancellation (ANC) and a live translation feature. You’ll be able to preorder the AirPods Pro 3 today for $249, and they’ll start shipping on September 19th.

Apple is upgrading the audio quality in the AirPods Pro 3 with a widened sound stage and improved noise cancellation. Apple says the AirPods Pro 3 will deliver the “world’s best ANC,” thanks to foam-infused ear tips for greater noise isolation. Apple claims this will enable twice the ANC than the previous generation of AirPods Pro, and “deliver the world’s best ANC of any in-ear wireless headphones.”

Perhaps the biggest upgrade on the AirPods Pro 3 will be the new heart rate sensor. Apple is using its smallest heart rate sensor on the AirPods Pro 3, which is a custom photoplethysmography (PPG) sensor that “shines invisible infrared light pulsed at 256 times per second to measure light absorption in blood flow.”

«

Ah, so it finally came true.

I’m more interested by this:

»

Apple is also bringing a new live translation feature to the AirPods Pro 3. ANC will lower the volume of the speaker and play audio back to you in your preferred language. “When enabled, Live Translation helps users understand another language and communicate with others by speaking naturally with AirPods,” explains Apple. “To interact with someone who doesn’t have this hands-free capability, there’s an option to use iPhone as a horizontal display, showing the live transcription of what the user is saying in the other person’s preferred language.”

«

I can’t honestly imagine being in a situation where I’d want to do this, and you have to upgrade the phone to iOS 26 (🤢), but it’s honestly the most attractive thing Apple announced in the entire infomercial. (Though Google announced it for its Pixel Buds back in November 2023. Do people use that?)
unique link to this extract

---

Zuckerberg’s AI hires disrupt Meta with swift exits and threats to leave • Financial Times

Hannah Murphy, Cristina Criddle and George Hammond:

»

“There’s a lot of big men on campus,” said one investor who is close with some of Meta’s new AI leaders. 

Adding to the tumult, a handful of new AI staff have already decided to leave after brief tenures, according to people familiar with the matter.

This includes Ethan Knight, a machine-learning scientist who joined the company weeks ago. Another, Avi Verma, a former OpenAI researcher, went through Meta’s onboarding process but never showed up for his first day, according to a person familiar with the matter.

In a tweet on X on Wednesday, Rishabh Agarwal, a research scientist who started at Meta in April, announced his departure. He said that while Zuckerberg and Wang’s pitch was “incredibly compelling”, he “felt the pull to take on a different kind of risk”, without giving more detail.

Meanwhile, Chaya Nayak and Loredana Crisan, generative AI staffers who had worked at Meta for nine and 10 years respectively, are among the more than half a dozen veteran employees to announce they are leaving in recent days. Wired first reported some details of recent exits, including Zhao’s threatened departure. 

Meta said: “We appreciate that there’s outsized interest in seemingly every minute detail of our AI efforts, no matter how inconsequential or mundane, but we’re just focused on doing the work to deliver personal superintelligence.”

«

Late to this (it was reported at the end of August). That’s a lot of money not quite taken up. (Via Benedict Evans, like quite a few of the links below.)
unique link to this extract

---

China’s social media platforms rush to abide by AI-generated content labelling law • South China Morning Post

Coco Feng:

»

Major Chinese social media platforms, including Tencent Holdings’ WeChat and ByteDance-owned Douyin, have launched new features to abide by Monday’s roll-out of a new law that mandates labelling of all artificial intelligence-generated content online.

The law, which was issued in March, requires explicit and implicit labels for AI-generated text, images, audio, video and other virtual content. Explicit markings must be clearly visible to users, while implicit identifiers – such as digital watermarks – should be embedded in the metadata.

The country’s top internet watchdog, the Cyberspace Administration of China (CAC) – along with the Ministry of Industry and Information Technology, the Ministry of Public Security and the National Radio and Television Administration – drafted the law.

The new regulation reflects Beijing’s increased scrutiny of AI, as concerns grow over misinformation, copyright infringement and online fraud.

«

Have to feel that it’s a strange state of affairs when the Chinese Communist Party is more concerned about copyright and fraud than the US Republican Party. The “misinformation” stuff, well, you can understand that bit.
unique link to this extract

---

Some French publishers are giving AI revenue directly to journalists. Could that ever happen in the US? • Nieman Journalism Lab

Andrew Deck:

»

In the US, when a publisher signs a licensing deal with an AI company, newsroom staffers don’t get a cut.

Many newsrooms have licensed their content to OpenAI in bulk, for example. A staff reporter’s stories can be used as training data for the latest GPT model, or may surface in ChatGPT’s response to a user question.

Does that reporter deserve to be compensated directly for how their work is being used by OpenAI? In France, the answer is, increasingly, yes.

At least, that’s the logic underlying a host of agreements between French news publishers and trade unions, which are redistributing revenue from AI licensing deals directly to journalists. These agreements guarantee that if a reporter’s stories are being used by AI companies, they will share directly in the publisher’s earnings.

In some cases, the agreements put a fixed sum of several hundred euros into the pockets of newsroom staffers each year. At other outlets, journalists are splitting a percentage share. Le Monde, one of France’s largest newspapers, signed a deal with several unions in June 2024 ensuring a quarter of its AI licensing revenue is redistributed, without a ceiling.

To an American journalist, these arrangements can feel foreign or out of reach.

«

I’ll go with “about as out of reach as the Moon”, actually. Any US publisher large enough to warrant some sort of payout is not going to pass it on. However as the article explains the French have a slightly different system for assigning rights.
unique link to this extract

---

Warner Bros. sues Midjourney to stop AI knockoffs of Batman, Scooby-Doo • Ars Technica

Ashley Belanger:

»

Warner Bros. hit Midjourney with a lawsuit last Thursday, crafting a complaint that strives to shoot down defenses that the AI company has already raised in a similar lawsuit filed by Disney and Universal Studios earlier this year.

The big film studios have alleged that Midjourney profits off image generation models trained to produce outputs of popular characters. For Disney and Universal, intellectual property rights to pop icons like Darth Vader and the Simpsons were allegedly infringed. And now, the WB complaint defends rights over comic characters like Superman, Wonder Woman, and Batman, as well as characters considered “pillars of pop culture with a lasting impact on generations,” like Scooby-Doo and Bugs Bunny, and modern cartoon characters like Rick and Morty.

“Midjourney brazenly dispenses Warner Bros. Discovery’s intellectual property as if it were its own,” the WB complaint said, accusing Midjourney of allowing subscribers to “pick iconic” copyrighted characters and generate them in “every imaginable scene.”

Planning to seize Midjourney’s profits from allegedly using beloved characters to promote its service, Warner Bros. described Midjourney as “defiant and undeterred” by the Disney/Universal lawsuit.

«

I suspect Midjourney is more like tired and weary of the lawsuits. The new process for an AI startup is: 1) create AI project 2) hire lawyers 3) ??? 4) perhaps profit, who knows?
unique link to this extract

---

• Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more.

---

Errata, corrigenda and ai no corrida: none notified