U.S. Treasury Secretary Janet Yellen will warn that the use of AI in finance could lower transaction costs, but carries "significant risks," according to excerpts from a speech to be delivered on Thursday. From a report: In the remarks to a Financial Stability Oversight Council and Brookings Institution AI conference, Yellen says AI-related risks have moved towards the top of the regulatory council's agenda.

"Specific vulnerabilities may arise from the complexity and opacity of AI models, inadequate risk management frameworks to account for AI risks and interconnections that emerge as many market participants rely on the same data and models," Yellen says in the excerpts. She also notes that concentration among the vendors that develop AI models and that provide data and cloud services may also introduce risks that could amplify existing third-party service provider risks. "And insufficient or faulty data could also perpetuate or introduce new biases in financial decision-making," according to Yellen.

Microsoft is ending support for Windows 10 in October 2025, but the company is now taking the unusual step of reopening its beta program for Windows 10 to test new features and improvements. From a report: Windows 10 already got the AI Copilot feature that was originally exclusive to Windows 11, and it may well get other features soon. "To bring new features and more improvements to Windows 10 as needed, we need a place to do active feature development with Windows Insiders," explains Microsoft's Windows Insider team in a blog post. "So today, we are opening the Beta Channel for Windows Insiders who are currently on Windows 10."

Microsoft hasn't revealed what additional Windows 10 features it plans to test next, but Windows Insiders can opt into the beta channel to get them early. Crucially, the Windows 10 end of support date of October 14th, 2025 is still unchanged. "Joining the Beta Channel on your Windows 10 PC does not change that," says Microsoft.

Apple is secretly developing robotic devices powered by generative AI, including a table-top robotic arm with an iPad-like display and a mobile robot for household chores, Bloomberg News is reporting, citing people familiar with the matter.

An anonymous reader quotes a report from Wired: When Microsoft CEO Satya Nadella revealed the new Windows AI tool that can answer questions about your web browsing and laptop use, he said one of the"magical" things about it was that the data doesn't leave your laptop; theWindows Recall system takes screenshots of your activity every five seconds and saves them on the device. But security experts say that data may not stay there for long. Two weeks ahead ofRecall's launch on new Copilot+ PCs on June 18, security researchers have demonstrated how preview versions of the tool store the screenshots in an unencrypted database. The researchers say the data could easily be hoovered up by an attacker. And now, in a warning about how Recall could be abused by criminal hackers, Alex Hagenah, a cybersecurity strategist and ethical hacker, has released a demo tool that can automatically extract and display everything Recall records on a laptop.

Dubbed TotalRecall -- yes, after the 1990 sci-fi film -- the tool can pull all the information that Recall saves into its main database on a Windows laptop. "The database is unencrypted. It's all plain text," Hagenah says. Since Microsoft revealed Recall in mid-May, security researchers have repeatedly compared it to spyware or stalkerware that can track everything you do on your device. "It's a Trojan 2.0 really, built in," Hagenah says, adding that he built TotalRecall -- which he's releasing on GitHub -- in order to show what is possible and to encourage Microsoft to make changes before Recall fully launches. [...] TotalRecall, Hagenah says, can automatically work out where the Recall database is on a laptop and then make a copy of the file, parsing all the data as it does so. While Microsoft's new Copilot+ PCs aren't out yet, it's possible to use Recall by emulating a version of the devices. "It does everything automatically," he says. The system can set a date range for extracting the data -- for instance, pulling information from only one specific week or day. Pulling one day of screenshots from Recall, which stores its information in an SQLite database, took two seconds at most, Hagenah says.

Included in what the database captures are screenshots of whatever is on your desktop -- a potential gold mine for criminal hackers or domestic abusers who may physically access their victim's device. Images include captures of messages sent on encrypted messaging apps Signal and WhatsApp, and remain in the captures regardless of whether disappearing messages are turned on in the apps. There are records of websites visited and every bit of text displayed on the PC. Once TotalRecall has been deployed, it will generate a summary about the data; it is also possible to search for specific terms in the database. Hagenah says an attacker could get a huge amount of information about their target, including insights into their emails, personal conversations, and any sensitive information that's captured by Recall. Hagenah's work builds on findings from cybersecurity researcher Kevin Beaumont, who has detailed how much information Recall captures and how easy it can be to extract it.

Today, Raspberry Pi introduced a new kit that adds AI functionality to the Raspberry Pi 5. ZDNet reports: The Raspberry Pi AI kit combines an M.2-format Hailo 8L AI accelerator with the Raspberry Pi M.2 HAT+ to create a powerful yet power-efficient solution. The Hailo-8L NPU (Neural Processing Unit) chip, capable of 13 trillion operations per second (TOPS), is built into an M.2 2242 form factor module that attaches to the M.2 HAT+. When connected to a Raspberry Pi 5 board running the latest Raspberry Pi OS, the NPU is automatically available for AI computing tasks. The AI module also has direct access to the Raspberry Pi's camera software stack and works with both first-party and third-party cameras.

The NPU allows the Raspberry Pi 5 to perform AI tasks such as object and facial recognition, human pose analysis, and more. Using an NPU frees up the Raspberry Pi 5's CPU, allowing it to focus on other tasks, making your projects more efficient and powerful. The Raspberry Pi AI kit is also compatible with the Raspberry Pi Active Cooler, ensuring optimal performance without overheating. Additionally, you can purchase a clear protective layer to prevent damage to the board, giving you peace of mind while working on your projects. The AI kit is priced at $70. It's available from Raspberry Pi Approved Resellers, including PiHut, PiShop.us, and CanaKit.

Sarah Perez reports via TechCrunch: After a multi-hour outage that took place in the early hours of the morning, OpenAI's ChatGPT chatbot went down again -- but this time, it wasn't the only AI provider affected. On Tuesday morning, both Anthropic's Claude and Perplexity began seeing issues, too, but these were more quickly resolved. Google's Gemini appears to be operating at present, though it may have also briefly gone offline, according to some user reports.

It's unusual for three major AI providers to all be down at the same time, which could signal a broader infrastructure issue or internet-scale problem, such as those that affect multiple social media sites simultaneously, for example. It's also possible that Claude and Perplexity's issues were not due to bugs or other issues, but from receiving too much traffic in a short period of time due to ChatGPT's outage.

A donation by former Google chief Eric Schmidt to Europe's top particle physics lab heralds a new way to fund frontier research just as the West's technological race with China quickens. From a report: The European Organization for Nuclear Research, or CERN, will use the previously unreported gift of $48 million [non-paywalled link] from the Eric & Wendy Schmidt Fund for Strategic Innovation to develop AI algorithms to analyze raw data from the lab's Large Hadron Collider, the world's most powerful energy particle accelerator. In 2012, it discovered the Higgs Boson, a particle that's key to understanding how the universe is built.

Now, CERN needs to reinvest to stay at the cutting edge of particle physics research. By the late 2030s, the LHC is expected to reach the end of its useful life and CERN needs $17 billion from European nations to fund the construction of a much bigger accelerator, known as the Future Circular Collider. But that funding has yet to be secured and, in the meantime, China has proposed its own collider. raditionally, CERN has relied on contributions from its 23 member states and observer partners like the US for funding pure research, while private investors focus on applied research, according to Charlotte Warakaulle, CERN's director of international relations. That makes the Schmidts' donation to pure research a private-sector first and may herald a different approach to funding the next collider, she says. "We're looking at all sorts of potential partners," Warakaulle said in an interview with Bloomberg last week. "How we could partner with the EU, private investments potentially."

Intel Chief Executive Officer Pat Gelsinger took the stage at the Computex show in Taiwan to talk about new products he expects will help turn back the tide of share losses to peers, including AI leader Nvidia. From a report: Intel showed its new Xeon 6 data center processors with more efficient cores that will allow operators to cut down the space required for a given task to a third of prior-generation hardware. Like rivals, from Advanced Micro Devices to Qualcomm, Intel touted benchmarks that showed its new silicon is significantly better than its existing options. AMD and Qualcomm's CEOs, in earlier Computex keynotes, used Intel's laptop and desktop processors to show how far ahead they are in certain aspects of technology.

Gelsinger took a direct shot at Nvidia CEO Jensen Huang's claim that traditional processors like Intel's are running out of steam in the age of artificial intelligence. "Unlike what Jensen would have you believe, Moore's Law is alive and well," he said, stressing that Intel will have a major role to play in the proliferation of AI as the leading provider of PC chips. "I think of it like the internet 25 years ago, it's that big," Gelsinger said. "We see this as the fuel that's driving the semiconductor industry to reach $1 trillion by the end of the decade."

A group of current and former employees from OpenAI and Google DeepMind are calling for protection from retaliation for sharing concerns about the "serious risks" of the technologies these and other companies are building. From a report: "So long as there is no effective government oversight of these corporations, current and former employees are among the few people who can hold them accountable to the public," according to a public letter, which was signed by 13 people who've worked at the companies, seven of whom included their names. "Yet broad confidentiality agreements block us from voicing our concerns, except to the very companies that may be failing to address these issues."

In recent weeks, OpenAI has faced controversy about its approach to safeguarding artificial intelligence after dissolving one of its most high-profile safety teams and being hit by a series of staff departures. OpenAI employees have also raised concerns that staffers were asked to sign nondisparagement agreements tied to their shares in the company, potentially causing them to lose out on lucrative equity deals if they speak out against the AI startup. After some pushback, OpenAI said it would release past employees from the agreements.

AMD is reverting to a simpler, more traditional numbering scheme for its laptop processors, abandoning its recent complex "decoder ring" system. The new system for Ryzen AI laptop processors will use a three-digit model number to denote generation and SKU, aligning more closely with industry norms. Ars Technica reports: For its new Ryzen AI laptop processors, codenamed "Strix Point," AMD is still using the same broad Ryzen 3/5/7/9 number to communicate general performance level plus a one- or two-letter suffix to denote general performance and power level (U for ultraportables, HX for higher-performance chips, and so on). A new three-digit processor number will inform buyers of the chip's generation in the first digit and denote the specific SKU using the last two digits. In other words, the company is essentially hitting the undo button.

Like Intel, AMD is shifting from four-digit numbers to three digits. The Strix Point processor numbers will start with the 300 series, which AMD says is because this is the third generation of Ryzen laptop processors with a neural processing unit (NPU) included. Current 7040-series and 8040-series processors with NPUs are not being renamed retroactively, and AMD plans to stop using the 7000- and 8000-series numbering for processor introductions going forward. AMD wouldn't describe exactly how it would approach CPU model numbers for new products that used older architectures but did say that new processors that didn't meet the 40+ TOPS requirement for Microsoft's Copilot+ program would simply use the "Ryzen" name instead of the new "Ryzen AI" branding. That would include older architectures with slower NPUs, like the current 7040 and 8040-series chips.

Desktop CPUs are, once again, totally unaffected by this change. Desktop processors' four-digit model numbers and alphabetic suffixes generally tell you all you need to know about their underlying architecture; the new Ryzen 9000 desktop CPUs and the Zen 5 architecture were also announced today. It seems like a lot of work to do to end up basically where we started, especially when the people at AMD who make and market the desktop chips have been getting by just fine with older model numbers for newly released products when appropriate. But to be fair to AMD, there just isn't a great way to do processor model numbers in a simple and consistent way, at least not given current market realities [...].

The Ansel Adams estate said it was "officially on our last nerve" after Adobe was caught selling AI-generated images imitating the late photographer's work. The Verge reports: While Adobe permits AI-generated images to be hosted and sold on its stock image platform, users are required to hold the appropriate rights or ownership over the content they upload. Adobe Stock's Contributor Terms specifically prohibits content "created using prompts containing other artist names, or created using prompts otherwise intended to copy another artist." Adobe responded to the callout, saying it had removed the offending content and had privately messaged the Adams estate to get in touch directly in the future. The Adams estate, however, said it had contacted Adobe directly multiple times since August 2023.

"Assuming you want to be taken seriously re: your purported commitment to ethical, responsible AI, while demonstrating respect for the creative community, we invite you to become proactive about complaints like ours, & to stop putting the onus on individual artists/artists' estates to continuously police our IP on your platform, on your terms," said the Adams estate on Threads. "It's past time to stop wasting resources that don't belong to you."

Adobe Stock Vice President Matthew Smith previously told The Verge that the company generally moderates all "crowdsourced" Adobe Stock assets before they are made available to customers, employing a "variety" of methods that include "an experienced team of moderators who review submissions." As of January 2024, Smith said the strongest action the company can take to enforce its platform rules is to block Adobe Stock users who violate them. Bassil Elkadi, Adobe's Director of Communications and Public Relations, told The Verge that Adobe is "actively in touch with Ansel Adams on this matter," and that "appropriate steps were taken given the user violated Stock terms." The Adams estate has since thanked Adobe for removing the images, and said that it expects "it will stick this time." "We don't have a problem with anyone taking inspiration from Ansel's photography," said the Adams estate. "But we strenuously object to the unauthorized use of his name to sell products of any kind, including digital products, and this includes AI-generated output -- regardless of whether his name has been used on the input side, or whether a given model has been trained on his work."

The next "Spider-Verse" film may have a new animation style: AI. From a report: Sony Pictures Entertainment (SPE) CEO Tony Vinciquerra does not mince words when it comes to AI. He likes the tech -- or at the very least, he likes the economics. "We are very focused on AI. The biggest problem with making films today is the expense," Vinciquerra said at Sony's Thursday (Friday in Japan) investor event. "We will be looking at ways to...produce both films for theaters and television in a more efficient way, using AI primarily." That's about the strongest support for AI we've heard from a film studio head.

"We had an 8-month strike over AI last year," Vinciquerra began his response to the first analyst question (from Nomura Securities) during his Q&A portion of the annual event. He also acknowledged that ongoing IATSE talks and the forthcoming Teamsters negotiations are "both over AI again." The sum total of those discussions between Hollywood's workers and its studios will inform just how far Vinciquerra and others can go. "The agreements that came out of last year's strikes and the agreements that come out of the IATSE and Teamsters [negotiations] will define roughly what we can do with AI," Vinciquerra said.

Zoom's CEO Eric Yuan predicts that AI will significantly transform the workplace, potentially ushering in a four-day workweek, he told The Verge in an interview. Yuan said Zoom is transitioning from a videoconferencing platform to a comprehensive collaboration suite called Zoom Workplace. He believes AI will automate routine tasks such as attending meetings, reading emails, and making phone calls, enabling employees to dedicate time to more creative and meaningful work. The Verge adds: The Verge: I'm asking you which meetings do you look at and think you would hand off?
Yuan: I started with the problem first, right? And last but not least, after the meeting is over, let's say I'm very busy and missed the meeting. I really don't understand what happened. That's one thing. Another thing for a very important meeting I missed, given I'm the CEO, they're probably going to postpone the meeting. The reason why is I probably need to make a decision. Given that I'm not there, they cannot move forward, so they have to reschedule. You look at all those problems. Let's assume AI is there. AI can understand my entire calendar, understand the context. Say you and I have a meeting -- just one click, and within five seconds, AI has already scheduled a meeting.

At the same time, every morning I wake up, an AI will tell me, "Eric, you have five meetings scheduled today. You do not need to join four of the five. You only need to join one. You can send a digital version of yourself." For the one meeting I join, after the meeting is over, I can get all the summary and send it to the people who couldn't make it. I can make a better decision. Again, I can leverage the AI as my assistant and give me all kinds of input, just more than myself. That's the vision.

British chip designer Arm expects to capture more than half of the Windows PC market within the next five years, CEO Rene Haas said in an interview. The company's optimism comes as Microsoft and its hardware partners gear up to introduce a new generation of AI-powered PCs running on Arm-designed chips, potentially reshaping the Intel-dominated industry. Haas attributed Microsoft's commitment to supporting Arm's technology through enhanced developer tools as a key factor in the anticipated market shift.

The highlight of AMD's presentation Sunday at Computex 2024 was "the introduction of AMD's Ryzen AI 300 Series processors for laptops and the Ryzen 9000 Series for desktops," writes Slashdot reader BrianFagioli (sharing his report at Beta News): AMD's Ryzen AI 300 Series processors, designed for next-generation AI laptops, come with AMD's latest XDNA 2 architecture. This includes a Neural Processing Unit (NPU) that delivers 50 TOPS of AI processing power, significantly enhancing the AI capabilities of laptops. Among the processors announced were the Ryzen AI 9 HX 370, which features 12 cores and 24 threads with a boost frequency of 5.1 GHz, and the Ryzen AI 9 365 with 10 cores and 20 threads, boosting up to 5.0 GHz...

In the desktop segment, the Ryzen 9000 Series processors, based on the "Zen 5" architecture, demonstrated an average 16% improvement in IPC performance over their predecessors built on the "Zen 4" architecture. The Ryzen 9 9950X stands out with 16 cores and 32 threads, reaching up to 5.7 GHz boost frequency and equipped with 80MB of cache... AMD also reaffirmed its commitment to the AM4 platform by introducing the Ryzen 9 5900XT and Ryzen 7 5800XT processors. These models are compatible with existing AM4 motherboards, providing an economical upgrade path for users.
The article adds that AMD also unveiled its Radeon PRO W7900 Dual Slot workstation graphics card — priced at $3,499 — "further broadening its impact on high-performance computing...

"AMD also emphasized its strategic partnerships with leading OEMs such as Acer, ASUS, HP, Lenovo, and MSI, who are set to launch systems powered by these new AMD processors." And there's also a software collaboration with Microsoft, reportedly "to enhance the capabilities of AI PCs, thus underscoring AMD's holistic approach to integrating AI into everyday computing."

AI models can evaluate how similar voices are to each other. So NPR asked forensic voice experts at Arizona State University to compare the voice and speech patterns of OpenAI's "Sky" to Scarlett Johansson's... The researchers measured Sky, based on audio from demos OpenAI delivered last week, against the voices of around 600 professional actresses. They found that Johansson's voice is more similar to Sky than 98% of the other actresses.

Yet she wasn't always the top hit in the multiple AI models that scanned the Sky voice. The researchers found that Sky was also reminiscent of other Hollywood stars, including Anne Hathaway and Keri Russell. The analysis of Sky often rated Hathaway and Russell as being even more similar to the AI than Johansson.

The lab study shows that the voices of Sky and Johansson have undeniable commonalities — something many listeners believed, and that now can be supported by statistical evidence, according to Arizona State University computer scientist Visar Berisha, who led the voice analysis in the school's College of Health Solutions and the College of Engineering. "Our analysis shows that the two voices are similar but likely not identical," Berisha said...

OpenAI maintains that Sky was not created with Johansson in mind, saying it was never meant to mimic the famous actress. "It's not her voice. It's not supposed to be. I'm sorry for the confusion. Clearly you think it is," Altman said at a conference this week. He said whether one voice is really similar to another will always be the subject of debate.

Slashdot reader storagedude shares a provocative post from the cybersecurity news blog of Cyble Inc. (a Ycombinator-backed company promising "AI-powered actionable threat intelligence").

The post delves into concerns that the new "Recall" feature planned for Windows (on upcoming Copilot+ PCs) is "a security and privacy nightmare." Copilot Recall will be enabled by default and will capture frequent screenshots, or "snapshots," of a user's activity and store them in a local database tied to the user account. The potential for exposure of personal and sensitive data through the new feature has alarmed security and privacy advocates and even sparked a UK inquiry into the issue. In a long Mastodon thread on the new feature, Windows security researcher Kevin Beaumont wrote, "I'm not being hyperbolic when I say this is the dumbest cybersecurity move in a decade. Good luck to my parents safely using their PC."

In a blog post on Recall security and privacy, Microsoft said that processing and storage are done only on the local device and encrypted, but even Microsoft's own explanations raise concerns: "Note that Recall does not perform content moderation. It will not hide information such as passwords or financial account numbers. That data may be in snapshots that are stored on your device, especially when sites do not follow standard internet protocols like cloaking password entry." Security and privacy advocates take issue with assertions that the data is stored securely on the local device. If someone has a user's password or if a court orders that data be turned over for legal or law enforcement purposes, the amount of data exposed could be much greater with Recall than would otherwise be exposed... And hackers, malware and infostealers will have access to vastly more data than they would without Recall.

Beaumont said the screenshots are stored in a SQLite database, "and you can access it as the user including programmatically. It 100% does not need physical access and can be stolen.... Recall enables threat actors to automate scraping everything you've ever looked at within seconds."
Beaumont's LinkedIn profile and blog say that starting in 2020 he worked at Microsoft for nearly a year as a senior threat intelligence analyst. And now Beaumont's Mastodon post is also raising other concerns (according to Cyble's blog post):

• "Sensitive data deleted by users will still be saved in Recall screenshots... 'If you or a friend use disappearing messages in WhatsApp, Signal etc, it is recorded regardless.'"

• "Beaumont also questioned Microsoft's assertion that all this is done locally."

The blog post also notes that Leslie Carhart, Director of Incident Response at Dragos, had this reaction to Beaumont's post. "The outrage and disbelief are warranted."

'"As AI programs shake up the office, potentially making millions of jobs obsolete, one group of perpetually stressed workers seems especially vulnerable..." writes the New York Times.

"The chief executive is increasingly imperiled by A.I." These employees analyze new markets and discern trends, both tasks a computer could do more efficiently. They spend much of their time communicating with colleagues, a laborious activity that is being automated with voice and image generators. Sometimes they must make difficult decisions — and who is better at being dispassionate than a machine?

Finally, these jobs are very well paid, which means the cost savings of eliminating them is considerable...

This is not just a prediction. A few successful companies have begun to publicly experiment with the notion of an A.I. leader, even if at the moment it might largely be a branding exercise... [The article gives the example of the Chinese online game company NetDragon Websoft, which has 5,000 employees, and the upscale Polish rum company Dictador.]

Chief executives themselves seem enthusiastic about the prospect — or maybe just fatalistic. EdX, the online learning platform created by administrators at Harvard and M.I.T. that is now a part of publicly traded 2U Inc., surveyed hundreds of chief executives and other executives last summer about the issue. Respondents were invited to take part and given what edX called "a small monetary incentive" to do so. The response was striking. Nearly half — 47 percent — of the executives surveyed said they believed "most" or "all" of the chief executive role should be completely automated or replaced by A.I. Even executives believe executives are superfluous in the late digital age...

The pandemic prepared people for this. Many office workers worked from home in 2020, and quite a few still do, at least several days a week. Communication with colleagues and executives is done through machines. It's just a small step to communicating with a machine that doesn't have a person at the other end of it. "Some people like the social aspects of having a human boss," said Phoebe V. Moore, professor of management and the futures of work at the University of Essex Business School. "But after Covid, many are also fine with not having one."
The article also notes that a 2017 survey of 1,000 British workers found 42% saying they'd be "comfortable" taking orders from a computer.

"Within less than a minute, I'm approached by a store worker who comes up to me and says, 'You're a thief, you need to leave the store'."

That's a quote from the BBC by a wrongly accused customer who was flagged by a facial-recognition system called Facewatch. "She says after her bag was searched she was led out of the shop, and told she was banned from all stores using the technology."

Facewatch later wrote to her and acknowledged it had made an error — but declined to comment on the incident in the BBC's report: [Facewatch] did say its technology helped to prevent crime and protect frontline workers. Home Bargains, too, declined to comment. It's not just retailers who are turning to the technology... [I]n east London, we joined the police as they positioned a modified white van on the high street. Cameras attached to its roof captured thousands of images of people's faces. If they matched people on a police watchlist, officers would speak to them and potentially arrest them...

On the day we were filming, the Metropolitan Police said they made six arrests with the assistance of the tech... The BBC spoke to several people approached by the police who confirmed that they had been correctly identified by the system — 192 arrests have been made so far this year as a result of it.
Lindsey Chiswick, director of intelligence for the Met, told the BBC that "It takes less than a second for the technology to create a biometric image of a person's face, assess it against the bespoke watchlist and automatically delete it when there is no match."

"That is the correct and acceptable way to do it," writes long-time Slashdot reader Baron_Yam, "without infringing unnecessarily on the freedoms of the average citizen. Just tell me they have appropriate rules, effective oversight, and a penalty system with teeth to catch and punish the inevitable violators."

But one critic of the tech complains to the BBC that everyone scanned automatically joins "a digital police line-up," while the article adds that others "liken the process to a supermarket checkout — where your face becomes a bar code." And "The error count is much higher once someone is actually flagged. One in 40 alerts so far this year has been a false positive..."

Thanks to Slashdot reader Bruce66423 for sharing the article.

How will Apple protect user data while their requests are being processed by AI in applications like Siri?

Long-time Slashdot reader AmiMoJo shared this report from Apple Insider: According to sources of The Information [four different former Apple employees who worked on the project], Apple intends to process data from AI applications inside a virtual black box.

The concept, known as "Apple Chips in Data Centers" internally, would involve only Apple's hardware being used to perform AI processing in the cloud. The idea is that it will control both the hardware and software on its servers, enabling it to design more secure systems. While on-device AI processing is highly private, the initiative could make cloud processing for Apple customers to be similarly secure... By taking control over how data is processed in the cloud, it would make it easier for Apple to implement processes to make a breach much harder to actually happen.

Furthermore, the black box approach would also prevent Apple itself from being able to see the data. As a byproduct, this means it would also be difficult for Apple to hand over any personal data from government or law enforcement data requests.
Processed data from the servers would be stored in Apple's "Secure Enclave" (where the iPhone stores biometric data, encryption keys and passwords), according to the article.

"Doing so means the data can't be seen by other elements of the system, nor Apple itself."