A group of Wikipedia editors have formed WikiProject AI Cleanup, "a collaboration to combat the increasing problem of unsourced, poorly-written AI-generated content on Wikipedia." From a report: The group's goal is to protect one of the world's largest repositories of information from the same kind of misleading AI-generated information that has plagued Google search results, books sold on Amazon, and academic journals. "A few of us had noticed the prevalence of unnatural writing that showed clear signs of being AI-generated, and we managed to replicate similar 'styles' using ChatGPT," Ilyas Lebleu, a founding member of WikiProject AI Cleanup, told me in an email. "Discovering some common AI catchphrases allowed us to quickly spot some of the most egregious examples of generated articles, which we quickly wanted to formalize into an organized project to compile our findings and techniques."

In many cases, WikiProject AI Cleanup finds AI-generated content on Wikipedia with the same methods others have used to find AI-generated content in scientific journals and Google Books, namely by searching for phrases commonly used by ChatGPT. One egregious example is this Wikipedia article about the Chester Mental Health Center, which in November of 2023 included the phrase "As of my last knowledge update in January 2022," referring to the last time the large language model was updated.

OpenAI's GPT Store, launched in January 2024, has failed to deliver on promised revenue-sharing for most small developers. Despite CEO Sam Altman's earlier statements about paying creators, only a select few have been invited to a pilot program, Wired is reporting.

Developers like Josh Brent Villocido, whose Books GPT was featured at launch, remain excluded from monetization opportunities. Many GPT creators report lack of analytics and unclear performance metrics. Some have devised workarounds, placing affiliate links or ads within their GPTs.

This week the U.S. Attorney's Office for the Southern District of New York unsealed charges over a "scheme to commit fraud" in carbon markets, which they say fraudulently netted one company "tens of millions of dollars" worth of credits — which led to "securing an investment of over $100 million."

MarketWatch reports: Ken Newcombe had spent years building a program to distribute more environmentally friendly cookstoves for free to rural communities in Africa and Southeast Asia. The benefit for his company, C-Quest Capital, would be the carbon credits it would receive in exchange for reducing the amount of fuel people burned in order to cook food — credits the company could then sell for a profit to big oil companies like BP.

But when Newcombe tried to ramp up the program, federal prosecutors said in an indictment made public Wednesday, he quickly realized that the stoves wouldn't deliver the emissions savings he had promised investors. Rather than admit his mistake, he and his partners cooked the books instead, prosecutors said... That allowed them to obtain carbon credits worth tens of millions of dollars that they didn't deserve, prosecutors said. On the basis of the fraudulently gained credits, prosecutors said, C-Quest was able to secure $250 million in funding from an outside investor.

"The alleged actions of the defendants and their co-conspirators risked undermining the integrity of [the global market for carbon credits], which is an important part of the fight against climate change," said Damian Williams, the U.S. attorney for the Southern District of New York.
From announced by the U.S. Attorney's Office: U.S. Attorney Damian Williams said... "The alleged actions of the defendants and their co-conspirators risked undermining the integrity of that market, which is an important part of the fight against climate change. Protecting the sanctity and integrity of the financial markets continues to be a cornerstone initiative for this Office, and we will continue to be vigilant in rooting out fraud in the market for carbon credits...."

While most carbon credits are created through, and trade in compliance markets, there is also a voluntary carbon market. Voluntary markets revolve around companies and entities that voluntarily set goals to reduce or offset their carbon emissions, often to align with goals from employees or shareholders. In voluntary markets, the credits are issued by non-governmental organizations, using standards for measuring emission reductions that they develop based on input from market participants, rather than on mandates from governments. The non-governmental organizations issue voluntary carbon credits to project developers that run projects that reduce emissions or remove greenhouse gases from the atmosphere.

CQC was a for-profit company that ran projects to generate carbon credits — including a type of credit known as a voluntary carbon unit ("VCU") — by reducing emissions of greenhouse gases. CQC profited by selling VCUs it obtained, often to companies seeking to offset the impact of greenhouse gases they emit in the course of operating their businesses.
The company itself was not charged due to "voluntary and timely self-disclosure of misconduct," according to the announcement, along with "full and proactive cooperation, timely and appropriate remediation, and agreement to cancel or void certain voluntary carbon units.

Novelist Christopher Farnsworth has filed a class-action lawsuit (PDF) against Meta, accusing the company of using his and other authors' pirated books to train its Llama AI model. Farnsworth seeks damages and an order to stop the alleged copyright infringement, joining a growing group of creators suing tech companies over unauthorized AI training. Reuters reports: Farnsworth said in the lawsuit on Tuesday that Meta fed Llama, which powers its AI chatbots, thousands of pirated books to teach it how to respond to human prompts. Other authors including Ta-Nehisi Coates, former Arkansas governor Mike Huckabee and comedian Sarah Silverman have brought similar class-action claims against Meta in the same court over its alleged use of their books in AI training. [...] Several groups of copyright owners including writers, visual artists and music publishers have sued major tech companies over the unauthorized use of their work to train generative AI systems. The companies have argued that their AI training is protected by the copyright doctrine of fair use and that the lawsuits threaten the burgeoning AI industry.

A U.S. Trademark Office tribunal canceled Marvel and DC's jointly owned "Super Hero" trademarks after the companies failed to respond to a request by London-based Superbabies Ltd, which argued the marks couldn't be owned collectively or monopolize the superhero genre. The ruling was "not just a win for our client but a victory for creativity and innovation," said Superbabies attorney Adam Adler of Reichman Jorgensen Lehman & Feldberg. "By establishing SUPER HEROES' place in the public domain, we safeguard it as a symbol of heroism available to all storytellers." Reuters reports: Rivals Marvel and DC jointly own four federal trademarks covering the terms "Super Hero" and "Super Heroes," the oldest of which dates back to 1967. Richold writes comics featuring a team of super-hero babies called the Super Babies. According to Richold, DC accused his company of infringing the "Super Hero" marks and threatened legal action after Superbabies Ltd applied for U.S. trademarks covering the "Super Babies" name. Marvel and DC have cited their marks in opposing dozens of superhero-related trademark applications at the USPTO, according to the office's records. Superbabies petitioned the office to cancel the marks in May. It argued that Marvel and DC cannot "claim ownership over an entire genre" with their trademarks, and that the two competitors cannot own trademarks together.

The creator of an open source project that scraped the internet to determine the ever-changing popularity of different words in human language usage says that they are sunsetting the project because generative AI spam has poisoned the internet to a level where the project no longer has any utility. 404 Media: Wordfreq is a program that tracked the ever-changing ways people used more than 40 different languages by analyzing millions of sources across Wikipedia, movie and TV subtitles, news articles, books, websites, Twitter, and Reddit. The system could be used to analyze changing language habits as slang and popular culture changed and language evolved, and was a resource for academics who study such things. In a note on the project's GitHub, creator Robyn Speer wrote that the project "will not be updated anymore."

"Generative AI has polluted the data," she wrote. "I don't think anyone has reliable information about post-2021 language usage by humans." She said that open web scraping was an important part of the project's data sources and "now the web at large is full of slop generated by large language models, written by no one to communicate nothing. Including this slop in the data skews the word frequencies." While there has always been spam on the internet and in the datasets that Wordfreq used, "it was manageable and often identifiable. Large language models generate text that masquerades as real language with intention behind it, even though there is none, and their output crops up everywhere," she wrote.

An anonymous reader quotes a report from Ars Technica: Earlier this year, we got a look at something unusual: the results of an internal investigation conducted by Harvard Business School that concluded one of its star faculty members had committed research misconduct. Normally, these reports are kept confidential, leaving questions regarding the methods and extent of data manipulations. But in this case, the report became public because the researcher had filed a lawsuit that alleged defamation on the part of the team of data detectives that had first identified potential cases of fabricated data, as well as Harvard Business School itself. Now, the court has ruled (PDF) on motions to dismiss the case. While the suit against Harvard will go on, the court has ruled that evidence-backed conclusions regarding fabricated data cannot constitute defamation -- which is probably a very good thing for science.

The researchers who had been sued, Uri Simonsohn, Leif Nelson, and Joe Simmons, run a blog called Data Colada where, among other things, they note cases of suspicious-looking data in the behavioral sciences. As we detailed in our earlier coverage, they published a series of blog posts describing an apparent case of fabricated data in four different papers published by the high-profile researcher Francesca Gino, a professor at Harvard Business School. The researchers also submitted the evidence to Harvard, which ran its own investigation that included interviewing the researchers involved and examining many of the original data files behind the paper. In the end, Harvard determined that research misconduct had been committed, placed Gino on administrative leave and considered revoking her tenure. Harvard contacted the journals where the papers were published to inform them that the underlying data was unreliable.

Gino then filed suit alleging that Harvard had breached their contract with her, defamed her, and interfered with her relationship with the publisher of her books. She also added defamation accusations against the Data Colada team. Both Harvard and the Data Colada collective filed a motion to have all the actions dismissed, which brings us to this new decision. Harvard got a mixed outcome. This appears to largely be the result that the Harvard Business School adopted a new and temporary policy for addressing research misconduct when the accusations against Gino came in. This, according to the court, leaves questions regarding whether the university had breached its contract with her. However, most of the rest of the suit was dismissed. The judge ruled that the university informing Gino's colleagues that Gino had been placed on administrative leave does not constitute defamation. Nor do the notices requesting retractions sent to the journals where the papers were published. "I find the Retraction Notices amount 'only to a statement of [Harvard Business School]'s evolving, subjective view or interpretation of its investigation into inaccuracies in certain [data] contained in the articles,' rather than defamation," the judge decided.

More critically, the researchers had every allegation against them thrown out. Here, the fact that the accusations involved evidence-based conclusions, and were presented with typical scientific caution, ended up protecting the researchers. The court cites precedent to note that "[s]cientific controversies must be settled by the methods of science rather than by the methods of litigation" and concludes that the material sent to Harvard "constitutes the Data Colada Defendants' subjective interpretation of the facts available to them." Since it had already been determined that Gino was a public figure due to her high-profile academic career, this does not rise to the standard of defamation. And, while the Data Colada team was pretty definitive in determining that data manipulation had taken place, its members were cautious about acknowledging that the evidence they had did not clearly indicate Gino was the one who had performed the manipulation. Finally, it was striking that the researchers had protected themselves by providing links to the data sources they'd used to draw their conclusions. The decision cites a precedent that indicates "by providing hyperlinks to the relevant information, the articles enable readers to review the underlying information for themselves and reach their own conclusions."

Hasbro CEO Chris Cocks revealed at a Goldman Sachs conference that the company has been using AI in game development, including for "Dungeons & Dragons" and "Magic: The Gathering," and plans to integrate AI further into gameplay, despite previously banning AI-generated content. "Inside of development, we've already been using AI," Cocks said. "It's mostly machine-learning-based AI or proprietary AI as opposed to a ChatGPT approach. We will deploy it significantly and liberally internally as both a knowledge worker aid and as a development aid." Futurism reports: While the logistical aspects of the technology seem fairly par for the course in the world of out-of-touch CEOs over-relying on it, Cocks then suggested that it will become a part of D&D gameplay. "I'm probably more excited though about the playful elements of AI," he said. "I play with probably 30 or 40 people regularly. There's not a single person who doesn't use AI somehow for either campaign development or character development or story ideas. That's a clear signal that we need to be embracing it."

After paying lip service to using AI "responsibly" and "paying creators for their work," Cocks then doubled down on his point. "The themes around using AI to enable user-generated content, using AI to streamline new player introduction, using AI for emergent storytelling -- I think you're going to see that not just our hardcore brands like D&D but also multiple of our brands," the Hasbro CEO said. Further reading: Magic: The Gathering Community Fears Generative AI Will Replace Talented Artists

Amazon's Audible will begin inviting a select group of US-based audiobook narrators to train AI on their voices, the clones of which can then be used to make audiobook recordings. From a report: The effort, which kicks off next week, is designed to add more audiobooks to the service, quickly and cheaply -- and to welcome traditional narrators into the evolving world of audiobook automation which, to date, many have regarded warily. Last year, Audible began offering US-based, self-published authors who make their books available on the Kindle Store the option of having their works narrated by a generic "virtual voice." The initiative has been popular. As of May, more than 40,000 books in Audible were marked as having made use of the technology. Under the new arrangement, rather than limiting the audio work entirely to company-owned synthetic voices, Audible will be encouraging professional narrators to get in on the action.

"Six years after sailing out of San Francisco with the ambition of developing the technology to rid the world's oceans of plastic, The Ocean Cleanup returned to San Francisco with the knowledge and know-how to relegate the Great Pacific Garbage Patch to the history books," according to a new announcement from the group.

As the Los Angeles Time describes it, "After three years extracting plastic waste from the notorious Great Pacific Garbage Patch, an environmental nonprofit says it can finish the job within a decade..." Twice the size of Texas, the mass of about 79,000 metric tons of plastic floating in the Pacific Ocean between California and Hawaii is growing at an exponential pace, according to researchers. At current levels, the cleanup would take a decade with a price tag of $7.5 billion, the Netherlands-based Ocean Cleanup said in a press release, announcing the group's intention to eliminate the garbage patch entirely. However, computer models suggest a more aggressive approach could complete the job in just five years and cost $4 billion.

The cleanup vessels deploy enormous u-shaped floating barriers to funnel trash toward a focal point where it can then be loaded aboard and brought to shore... In their three years at sea, the Ocean Cleanup vessels have removed more than a million pounds of trash, representing 0.5% of the total accumulation. "We have shown the world that the impossible is now possible. The only missing thing is who will ensure this job gets done," said Boyan Slat, founder and chief executive of the Ocean Cleanup.
Project founder Boyan Slat said in their statement that "Today's announcement is clear: clean oceans can be achieved in a manageable time and for a clear cost.

"Through the hard work of the past 10 years, humanity has the tools needed to clean up the ocean. We have shown the world that the impossible is now possible... [F]or the first time, we can tell the world what it costs, what is needed and how long it could take. It is time for action."

Next year the group will take "a one-year operational hiatus," according to the announcement — to deploy a new initiative mapping areas of intense plastic accumulation to make extractions "more impactful."

A pair of devices launching Wednesday highlight a growing trend in consumer hardware: doing one thing well, Axios writes. From the report: The smartphone rendered many formerly standalone devices obsolete, but now some tech with a single purpose can offer an experience that a digital Swiss Army knife can't. GoPro announced its latest cameras on Wednesday. The $199 GoPro Hero is a smaller, simpler camera that presents fewer controls, while this year's flagship -- the $399 Hero 13 Black -- supports a series of add-on lenses without losing its core as a rugged, waterproof action camera.

Meanwhile, the reMarkable tablet company is adding its first color model. ReMarkable's tablets allow creators to write and sketch without the distractions of more full-featured devices. ReMarkable's paper tablets are similar to Amazon's Kindle Scribe, but with a focus on creating and editing documents rather than reading digital books. In addition to adding color -- the most requested feature -- the Paper Pro has built-in illumination, a larger eInk display and an active digital pen, which allows digital ink to appear on the display within an imperceptible 12 milliseconds. The new Paper Pro ($579 or $629, depending on which pen is bundled) is still aimed at those who want to avoid notifications rather than those who want to multitask.

Internet Archive's "controlled digital lending" system and removal of controls during the pandemic don't qualify as fair use, the Second Circuit affirmed Wednesday. Bloomberg Law: Four major book publishers again thwarted the online repository's defense that its one-to-one lending practices mirrored those of traditional libraries, this time at the US Court of Appeals for the Second Circuit. Copying books in their entirety isn't transformative, and lending them for free competes with the publishers own book and ebook offerings, the unanimous panel said. Internet Archive said in a statement: We are disappointed in today's opinion about the Internet Archive's digital lending of books that are available electronically elsewhere. We are reviewing the court's opinion and will continue to defend the rights of libraries to own, lend, and preserve books. Further reading: Full-text of court opinion [PDF].

Long-time Slashdot reader theodp writes: The Contingency Contingent, is Leigh Claire La Berge's amazing tale of what she calls her "fake job in Y2K preparedness." La Berge offers an insider's view of the madness that ensued when Y2K panic gave rise to seemingly-limitless spending at mega-corporations for massive enterprise-wide Y2K remediation projects led by management consulting firms that left clients with little to show for their money. (La Berge was an analyst for consulting firm Arthur Andersen, where "the Andersen position was that 'Y2K is a documentation problem, not a technology problem'.... At a certain point all that had happened yesterday was our documenting, so then we documented that. Then, exponentially, we had to document ourselves documenting our own documentation."). In what reads like the story treatment for an Office Space sequel, La Berge writes that it was a fake job "because Andersen was faking it."
From the article: The firm spent the late 1990s certifying fraudulent financial statements from Enron, the Texas-based energy company that made financial derivatives a household phrase, until that company went bankrupt in a cloud of scandal and suicide and Andersen was convicted of obstruction of justice, surrendered its accounting licenses, and shuttered. But that was later.

Finally, it was a fake job because the problem that the Conglomerate had hired Andersen to solve was not real, at least not in the sense that it needed to be solved or that Andersen could solve it. The problem was known variously as Y2K, or the Year 2000, or the Y2K Bug, and it prophesied that on January 1, 2000, computers the world over would be unable to process the thousandth-digit change from 19 to 20 as 1999 rolled into 2000 and would crash, taking with them whatever technology they were operating, from email to television to air-traffic control to, really, the entire technological infrastructure of global modernity. Hospitals might have emergency power generators to stave off the worst effects (unless the generators, too, succumbed to the Y2K Bug), but not advertising firms.

With a world-ending scenario on the horizon, employment standards were being relaxed. The end of the millennium had produced a tight labor market in knowledge workers, and new kinds of companies, called dot-coms, were angling to dominate the emergent world of e-commerce. Flush with cash, these companies were hoovering up any possessors of knowledge they could find. Friends from my gradeless college whose only experience in business had been parking-lot drug deals were talking stock options.
Looking back, the author remembers being "surprised by how quickly Y2K disappeared from office discourse as though censored..."

Their upcoming book is called Fake Work: How I Began to Suspect Capitalism is a Joke.

Apple has laid off about 100 employees in its services group (source may be paywalled; alternative source), primarily affecting roles associated with the Apple Books app and Apple Bookstore. The San Francisco Chronicle reports: The impacted employees at the Cupertino-based tech giant were informed of the cuts on Tuesday, Bloomberg reported (paywalled). The layoffs spanned various teams under Senior Vice President Eddy Cue. The job cuts include roles primarily associated with the Apple Books app and Apple Bookstore, with the company shifting its focus to other divisions. Additionally, other services teams, such as the one managing Apple News, also experienced layoffs.

While Apple has largely avoided mass layoffs even as other major tech companies have downsized, it did lay off 614 employees in Santa Clara earlier this year. Those cuts marked Apple's first significant job reductions since the onset of the COVID-19 pandemic and coincided with the cancellation of its decade-long electric car project.

An anonymous reader shared this story from the blog Decrypt: Michael Lewis, author of Going Infinite, an account of the rise and fall of Sam Bankman-Fried, has argued that the disgraced FTX founder didn't have "the character of a thief" in a new The Washington Post article. "His crime was of a piece with his character. The character wasn't the character of a thief. It was the character of a person numb to risk." Lewis explained in the final paragraphs of a 4,500 word essay adapted from a new introduction to his book. "Unable to feel risk himself, he can't really imagine other people feeling much at all about the risk he has subjected them to...."

Lewis doubled down on previous claims that Bankman-Fried wasn't running a Ponzi scheme, arguing that "The crime was unnecessary to the business in a way that, say, Bernie Madoff's was not," and that "The crime made no sense." The collapse of FTX, he added, "might have been avoided and FTX might have survived."

"That doesn't mean I think that Sam Bankman-Fried is innocent. It merely informs how I feel about him," Lewis explained. "I think the truth is closer to 'young person with an intellectually defensible but socially unacceptable moral code makes a huge mistake in trying to live by it' than "criminal on the loose in the financial system.'"
From from The Daily Beast: Lewis also pointed to bankruptcy court filings from FTX in the weeks after Bankman-Fried's sentencing showing that "against the $8.7 billion in missing customer deposits, FTX was now sitting on something like $14.5 to $16.3 billion." "Whatever the exact sum, it was enough to repay all depositors and various other creditors at least 118 cents on the dollar — that is, everyone who imagined they had lost money back in November 2022 would get their money back, with interest," Lewis writes.
Michael Lewis's article offers some vivid details: Inside of three years, he'd gone from socially and emotionally isolated 25-year-old with an upper-middle-class bank account to leader of a small army of math nerds and (according to Forbes magazine) not merely the world's richest person under 30 but maybe the fastest creator of wealth in recorded history... He'd gone from having no friends as a child to having too many as an adult without ever developing a capacity for friendship....

The prosecutors didn't need Sam's help. Sam helped them anyway by ignoring the counsel of his lawyers and testifying on his own behalf... As Lewis Kaplan, the federal judge who presided over the case, said later: "When he wasn't outright lying, he was often evasive, hairsplitting, dodging questions and trying to get the prosecutor to reword questions in ways that he could answer in ways he thought less harmful than a truthful answer to the question that was posed would have been. I've been doing this job for close to 30 years. I've never seen a performance quite like that...." [T]he judge ordered Sam to rise so that he might address him directly. Two hours or so earlier, Sam had shuffled into the courtroom in prison khakis with his head down and his hands oddly clasped behind his back. Just before he'd entered, his guards had told him he was meant to be wearing handcuffs and asked if he could create the impression that he was doing so...

"There is a risk that this man will be in a position to do something very bad in the future, and it's not a trivial risk, not a trivial risk at all," said the judge. "So, in part, my sentence will be for the purpose of disabling him." He then sentenced Sam to 25 years in prison, with no possibility of parole.

A few minutes later, Sam dutifully clasped his hands behind his back and shuffled out of the courtroom.
Lewis adapted his 4,500-word article from the upcoming (updated) paperback edition of his book — which was originally published in 2023 on the same day jurors were selected for Bankman-Fried's trial...