Salesforce CEO Marc Benioff said his firm, San Francisco's largest private employer, [1]does not plan to hire engineers this year because of the success of AI agents created and used by the company. From a report:

> "My message to CEOs right now is that we are the last generation to manage only humans," Benioff said Wednesday on Salesforce's earnings call, indicating that companies of the future will have hybrid human and digital workforces. Benioff added that Salesforce's mission is to become "the No. 1 digital labor provider, period" to other companies.



[1] https://sfstandard.com/2025/02/27/salesforce-marcbenioff-layoffs-tech-agents/

An anonymous reader quotes a report from TechCrunch:

> OpenAI intends to eventually integrate its AI video generation tool, Sora, [1]directly into its popular consumer chatbot app, ChatGPT , company leaders said during a Friday office hours session on Discord. Today, Sora is only available through a [2]dedicated web app OpenAI launched in December, which lets users access the AI video model of the same name to generate up to twenty-second-long cinematic clips. However, OpenAI's product lead for Sora, Rohan Sahai, said the company has plans to put Sora in more places, and expand what Sora can create.

>

> [...] OpenAI may be trying to attract users to ChatGPT by letting them generate Sora videos from the chatbot. Putting Sora in ChatGPT could also incentivize users to upgrade to ChatGPT's premium subscription tiers, which may offer higher video generation limits. One of the reasons OpenAI launched Sora as a separate web app was to maintain ChatGPT's simplicity, Sahai explained during the office hours. Since its launch, OpenAI has expanded Sora's web experience, creating more ways for users to browse Sora-generated videos from the community. Sahai also said OpenAI "would love to build" a standalone mobile app for Sora, noting that the Sora team is actively looking for mobile engineers.

OpenAI also plans to expand Sora's generation capabilities to images, letting users create more photorealistic images than what's currently possible with OpenAI's DALL-E3 model.



[1] https://techcrunch.com/2025/02/28/openai-plans-to-integrate-soras-video-generator-into-chatgpt/

[2] https://sora.com/library

The Ninth Circuit Court of Appeals is set to review a California district court's ruling in Neo4j v. PureThink, which upheld Neo4j's right to modify the GNU AGPLv3 with additional binding terms. If the appellate court affirms this decision, it could set a precedent [1]allowing licensors to impose unremovable restrictions on open-source software , potentially undermining the enforceability of GPL-based licenses and threatening the integrity of the open-source ecosystem. The Register reports:

> The GNU AGPLv3 is a free and open source software (FOSS) license [2]largely based on the [3]GNU GPLv3 , both of which are published by the Free Software Foundation (FSF). Neo4j provided database software under the AGPLv3, then tweaked the license, leading to legal battles over forks of the software. The AGPLv3 includes language that says any added restrictions or requirements are removable, meaning someone could just file off Neo4j's changes to the usage and distribution license, reverting it back to the standard AGPLv3, which the biz has argued and successfully fought against in that California district court.

>

> Now the matter, the validity of that modified FOSS license, is before an appeals court in the USA. "I don't think the community realizes that if the Ninth Circuit upholds the lower court's ruling, it won't just kill GPLv3," PureThink's John Mark Suhy told The Register. "It will create a dangerous legal precedent that could be used to undermine all open-source licenses, allowing licensors to impose unexpected restrictions and fundamentally eroding the trust that makes open source possible."

>

> Perhaps equally concerning is the fact that Suhy, founder and CTO of PureThink and iGov (the two firms sued by Neo4j), and presently CTO of IT consultancy Greystones Group, is defending GPL licenses on his own, pro se, without the help of the FSF, founded by Richard Stallman, creator of the GNU General Public License. "I'm actually doing everything pro se because I used up all my savings to fight it in the lower court," said Suhy. "I'm surprised the Free Software Foundation didn't care too much about it. They always had an excuse about not having the money for it. Luckily the Software Freedom Conservancy came in and helped out there."



[1] https://www.theregister.com/2025/02/27/adverse_appeals_court_ruling_could/

[2] https://www.gnu.org/licenses/why-affero-gpl.en.html

[3] https://www.gnu.org/licenses/gpl-3.0.en.html

According to The Columbia Dispatch, Intel's [1]promised $28 billion semiconductor project in central Ohio has been [2]delayed again until 2030 , with operations beginning sometime shortly thereafter in either 2030 or 2031. From the report:

> By the time it opens, Intel's first factory will have faced at least five or six years of delays, as it was originally scheduled to begin operating in 2025. Intel's second Ohio factory won't be completed until at least 2031 and will begin running in 2032, according to the company. The new timeline comes as Intel continues to struggle financially, which was a key factor in the latest delay for the company's Ohio factories. The company was alerting its employees of the delays in a message Friday.

>

> The changes were made so Intel can align its factory operation with market demand and better "manage our capital responsibly," Naga Chandrasekaran, executive vice president, chief global operations officer and general manager of Intel Foundry Manufacturing wrote in a message to workers. The changes will ensure Intel's Ohio fabs will be finished in a "financially responsible manner that sets up Ohio One for success," Chandrasekaran wrote. "I wanted to be upfront and transparent with you all about our current plan. In no way does this diminish our long-term commitment to Ohio," Chandrasekaran wrote. "(W)e will continue to scale our hiring as we approach our operational dates. Intel is proud to call Ohio home, and we remain excited about our future here."



[1] https://slashdot.org/story/22/01/21/1528229/intel-reveals-plans-for-massive-new-ohio-factory-fighting-the-chip-shortage-stateside?sdsrc=rel

[2] https://www.dispatch.com/story/news/2025/02/28/intel-ohio-chip-factory-delayed-new-albany/80732342007/

An anonymous reader quotes a report from Ars Technica:

> Thousands" of complaints about the volume of TV commercials have [1]flooded the Federal Communications Commission (FCC) in recent years . Despite the FCC requiring TV stations, cable operators, and satellite providers to ensure that commercials don't bring a sudden spike in decibels, complaints around loud commercials "took a troubling jump" in 2024, the government body said on Thursday.

>

> Under The Commercial Advertisement Loudness Mitigation (CALM) Act, broadcast, cable, and satellite TV providers are required to ensure that commercials "have the same average volume as the programs they accompany," per the [2]FCC . The FCC's rules about the volume of commercials took effect in December 2012. The law also requires linear TV providers to use the Advanced Television Systems Committee's (ATSC's) [3]recommended practices . The practices include guidance around production, post production, metadata systems usage, and controlling dynamic range. If followed, the recommendations "result in consistency in loudness and avoidance of signal clipping," per the ATSC [ [4]PDF ]. The guidance reads: "If all programs and commercials were produced at a consistent average loudness, and if the loudness of the mix is preserved through the production, distribution, and delivery chain, listeners would not be subjected to annoying changes in loudness within and between programs."

>

> As spotted by [5]PC Mag , the FCC claimed this week that The Calm Act initially reduced complaints about commercials aggressively blaring from TVs. However, the agency is seeing an uptick in grievances. The FCC said it received "approximately" 750 complaints in 2022, 825 in 2023, and "at least" 1,700 in 2024 [ [6]PDF ]. Since The Calm Act regulates a commercial's average loudness, some advertisers may be skirting the spirit of the law by making commercials very loud at the start (to get viewers' attention) before quieting down for the rest of the ad. In response to growing complaints, the FCC is reexamining its rules and this week announced that it's seeking comment from "consumers and industry on the extent to which The CALM Act rules are effective." The FCC is also asking people to weigh in on what future actions the FCC, the TV industry, or standard developers could take.

The FCC is considering whether to extend the Calm Act to online streaming services, which are increasingly offering plans with ad-supported models and live event broadcasts.



[1] https://arstechnica.com/gadgets/2025/02/commercials-are-still-too-loud-say-thousands-of-recent-fcc-complaints/

[2] https://www.fcc.gov/enforcement/areas/sound-volume-commercials-calm-act

[3] https://www.atsc.org/atsc-documents/a85-techniques-for-establishing-and-maintaining-audio-loudness-for-digital-television/

[4] https://cdn.arstechnica.net/wp-content/uploads/2025/02/A85-2013.pdf

[5] https://www.pcmag.com/news/are-tv-commercials-too-loud-fcc-to-investigate-eyes-streaming-services

[6] https://cdn.arstechnica.net/wp-content/uploads/2025/02/FCC.pdf

Mozilla has [1]denied allegations that its new Firefox browser terms of service allow it to [2]harvest user data for artificial intelligence training , following widespread criticism of the recently updated policy language. The controversy erupted after Firefox introduced terms that grant Mozilla "a nonexclusive, royalty-free, worldwide license to use that information" when users upload content through the browser, prompting competitor Brave Software's CEO Brendan Eich to suggest a business pivot toward data monetization.

"These changes are not driven by a desire by Mozilla to use people's data for AI or sell it to advertisers," Mozilla spokesperson Kenya Friend-Daniel told TechCrunch. "Our ability to use data is still limited by what we disclose in the Privacy Notice." The company clarified that its AI features operate locally on users' devices and don't send content data to Mozilla. Any data shared with advertisers is provided only on a "de-identified or aggregated basis," according to the spokesperson. Mozilla explained it used specific legal terms -- "nonexclusive," "royalty-free," and "worldwide" -- because Firefox is free, available globally, and allows users to maintain control of their own data.



[1] https://techcrunch.com/2025/02/28/mozilla-responds-to-backlash-over-new-terms-saying-its-not-using-peoples-data-for-ai/

[2] https://news.slashdot.org/story/25/02/28/1147239/mozillas-updated-tos-we-own-all-info-you-put-into-firefox

Google co-founder Sergey Brin has urged employees working on the company's Gemini AI products to be in the office " [1]at least every weekday " [ [2]non-paywalled source ] and suggested "60 hours a week is the sweet spot of productivity," according to an internal memo cited by The New York Times. The directive comes as Brin warned that "competition has accelerated immensely and the final race to A.G.I. is afoot," referring to artificial general intelligence, when machines match or surpass human intelligence.

"I think we have all the ingredients to win this race, but we are going to have to turbocharge our efforts," Brin wrote in the Wednesday evening memo. The guidance does not alter Google's official policy requiring employees to work in-office three days weekly. Brin, who returned to Google following ChatGPT's 2022 launch, also criticized staff who "put in the bare minimum," calling them "highly demoralizing to everyone else."



[1] https://www.nytimes.com/2025/02/27/technology/google-sergey-brin-return-to-office.html

[2] https://fortune.com/2025/02/28/sergey-brin-60-hours-work-week-rto-gemini-artificial-general-intelligence/

Workers reported [1]saving a substantial number of work hours by using generative AI, according to research conducted by the Federal Reserve Bank of St. Louis, along with Vanderbilt and Harvard universities. From a report:

> The researchers, drawing from what they identified as the first nationally representative survey of generative AI adoption, measured the impact of generative AI on work productivity by how much workers used the technology and how intensely. They found users are saving meaningful amounts of time.

>

> "On average, workers are 33% more productive in each hour that they use generative AI," the paper found. Among respondents that used generative AI in the previous week, 21% said it saved them four hours or more in that week, 20% reported three hours, 26% said two hours and 33% reported an hour or less.



[1] https://www.straitstimes.com/world/united-states/us-workers-see-ai-induced-productivity-growth-us-fed-survey-shows

Google DeepMind CEO Demis Hassabis says the definition of artificial general intelligence is being " [1]watered down ," creating an illusion of faster progress toward this technological milestone. "There's quite a long way, in my view, before we get to AGI," Hassabis said. "The timelines are shrinking because the definition of AGI is being watered down, in my opinion." DeepMind defines AGI as "AI systems that are at least as capable as humans at most cognitive tasks," while OpenAI has historically described it as a "highly autonomous system that outperforms humans at most economically valuable work."

OpenAI CEO Sam Altman recently declared his team is "confident we know how to build AGI," while modifying his personal definition to an AI "system that can tackle increasingly complex problems, at human level, in many fields." Hassabis suggested industry hype might be financially motivated: "There is a lot of hype for various reasons," he said, including perhaps "that people need to raise money." Microsoft CEO Satya Nadella separately dismissed AGI milestones as "nonsensical benchmark hacking," preferring economic impact measurements.



[1] https://www.bloomberg.com/news/newsletters/2025-02-27/why-tech-leaders-are-divided-on-a-key-milestone-for-ai

President Trump has directly criticized the UK government's [1]approach to encryption , comparing recent actions to those of China. Speaking to The Spectator, Trump said he confronted UK Prime Minister Keir Starmer about the Home Office's request for "backdoor access" to encrypted iCloud data, which led Apple to [2]remove its Advanced Data Protection feature from British services entirely.

"We told them you can't do this... That's incredible. That's something, you know, that you hear about with China," Trump [3]said after his meeting with Starmer . The remarks come as the Trump administration has directed Treasury and Commerce officials to examine UK tech regulations, including the Online Safety Act, for potential free speech violations and discrimination against US companies.



[1] https://apple.slashdot.org/story/25/02/07/1150200/uk-orders-apple-to-let-it-spy-on-users-encrypted-accounts

[2] https://apple.slashdot.org/story/25/02/21/1529255/apple-removes-cloud-encryption-feature-from-uk-after-backdoor-order

[3] https://www.spectator.co.uk/article/trump-uk-encryption-laws-are-like-what-you-get-in-china/

Microsoft said Friday it will [1]shut down its Skype messaging service on May 5 , replacing it with the free version of Microsoft Teams for consumers. Existing Skype users will have approximately 60 days to decide whether to migrate to Teams, where their message history, group chats and contacts will automatically transfer, or export their data including photos and conversation history.

The company will discontinue Skype's telephony features for calling domestic and international numbers, though it will honor existing Skype credits and subscriptions inside Teams until users' next renewal period. Skype Number users will need to port their numbers to other providers. Microsoft [2]acquired Skype for $8.5 billion in 2011 . The shutdown will not result in immediate job cuts.



[1] https://www.xda-developers.com/microsoft-killing-skype/

[2] https://news.slashdot.org/story/11/10/14/1157230/microsoft-finalizes-skype-acquisition

Google changed its rules around how product-review sites appear in its search engine. In the process, it [1]devastated a once-lucrative corner [ [2]non-paywalled source ] of the news media world. From a report:

> Sites including CNN Underscored and Forbes Vetted offer tips on everything from mattresses and knife sets to savings accounts, making money when users click on links and buy products.

>

> They depend on Google to drive much of their traffic, and therefore revenue. But over the past year, Google created stricter rules that dinged certain sites that farm out articles to freelancers, among other things. The goal, Google has said, was to give users higher-quality search results. The outcome was a crisis for some sites. Traffic for Forbes Advisor, a personal-finance recommendation site, fell 83% in January from the same month the year before, according to data firm Similarweb.

>

> CNN Underscored and Buy Side from WSJ, which is operated by Wall Street Journal parent Dow Jones, were both down by more than 25% in that period. Time magazine's Time Stamped and the Associated Press's AP Buyline, powered by Taboola Turnkey Commerce, ended their efforts in recent months. Taboola closed the commerce operation.



[1] https://www.wsj.com/business/media/google-search-change-product-recommendation-websites-02394b79

[2] https://www.emarketer.com/content/google-rule-change-hits-product-review-sites

Citigroup [1]credited a client's account with $81 trillion when it meant to send only $280, an error that could hinder the bank's attempt to persuade regulators that it has fixed long-standing operational issues. Financial Times:

> The erroneous internal transfer, which occurred last April and has not been previously reported, was missed by both a payments employee and a second official assigned to check the transaction before it was approved to be processed at the start of business the following day.

>

> A third employee detected a problem with the bank's account balances, catching the payment 90 minutes after it was posted. The payment was reversed several hours later, according to an internal account of the event seen by the Financial Times and two people familiar with the event. No funds left Citi, which disclosed the "near miss" to the Federal Reserve and Office of the Comptroller of the Currency, according to another person with knowledge of the matter.



[1] https://www.ft.com/content/9921925e-5a32-48cc-a3e3-3f77042477d2

New submitter [1]SharkByte writes:

> Mozilla just updated its Terms of Use and Privacy Policy for Firefox with a very disturbing "You Give Mozilla Certain Rights and Permissions" clause:

>

>

> When you upload or input information through Firefox, you hereby grant us a nonexclusive, royalty-free, worldwide license to use that information to help you navigate, experience, and interact with online content as you indicate with your use of Firefox.

H/T to reader [2]agristin as well, who also wrote about this.



[1] https://slashdot.org/~SharkByte

[2] https://slashdot.org/~agristin

Memory safety bugs are "eroding trust in technology and costing billions," [1]argues a new post on Google's security blog — adding that "traditional approaches, like code auditing, fuzzing, and exploit mitigations — while helpful — haven't been enough to stem the tide."

So the blog post calls for a "common framework" for "defining specific, measurable criteria for achieving different levels of memory safety assurance." The hope is this gives policy makers "the technical foundation to craft effective policy initiatives and incentives promoting memory safety" leading to "a market in which vendors are incentivized to invest in memory safety." ("Customers will be empowered to recognize, demand, and reward safety.")

In January the same Google security researchers helped co-write an article noting there are now [2]strong memory-safety "research technologies" that are sufficiently mature : memory-safe languages (including "safer language subsets like [3]Safe Buffers for C++"), mathematically rigorous formal verification, software compartmentalization, and hardware and software protections. (With hardware protections including things like ARM's [4]Memory Tagging Extension and the ( [5]Capability Hardware Enhanced RISC Instructions , or "CHERI", architecture.) Google's security researchers are now calling for "a blueprint for a memory-safe future" — though Importantly, the idea is "defining the desired outcomes rather than locking ourselves into specific technologies."

Their blog post this week again urges a practical/actionable framework that's commonly understood, but one that supports different approaches (and allowing tailoring to specific needs) while enabling objective assessment:

> At Google, we're not just advocating for standardization and a memory-safe future, we're actively working to build it. We are collaborating with industry and academic partners to develop potential standards, and our joint authorship of the recent [6]CACM call-to-action marks an important first step in this process... This commitment is also reflected in our internal efforts. We are prioritizing memory-safe languages, and have already seen significant reductions in vulnerabilities by adopting languages like Rust in combination with existing, wide-spread usage of Java, Kotlin, and Go where performance constraints permit. We recognize that a complete transition to those languages will take time. That's why we're also investing in techniques to improve the safety of our existing C++ codebase by design, such as [7]deploying hardened libc++ .

>

> This effort isn't about picking winners or dictating solutions. It's about creating a level playing field, empowering informed decision-making, and driving a virtuous cycle of security improvement... The journey towards memory safety requires a collective commitment to standardization. We need to build a future where memory safety is not an afterthought but a foundational principle, a future where the next generation inherits a digital world that is secure by design.

The security researchers' post calls for "a collective commitment" to eliminate memory-safety bugs, "anchored on secure-by-design practices..." One of the blog post's subheadings? "Let's build a memory-safe future together."

And they're urging changes "not just for ourselves but for the generations that follow."



[1] https://security.googleblog.com/2025/02/securing-tomorrows-software-need-for.html

[2] https://cacm.acm.org/opinion/it-is-time-to-standardize-principles-and-practices-for-software-memory-safety/

[3] https://clang.llvm.org/docs/SafeBuffers.html

[4] https://community.arm.com/arm-community-blogs/b/architectures-and-processors-blog/posts/enhancing-memory-safety

[5] https://en.wikipedia.org/wiki/Capability_Hardware_Enhanced_RISC_Instructions

[6] https://cacm.acm.org/opinion/it-is-time-to-standardize-principles-and-practices-for-software-memory-safety/

[7] https://security.googleblog.com/2024/11/retrofitting-spatial-safety-to-hundreds.html

[1]BrianFagioli shares a report from BetaNews:

> [T]he MTA is testing a system that effectively [2]transforms Google Pixel Android smartphones into powerful diagnostic tools for tracking rail defects . The project, called " [3]TrackInspect ," attaches Google Pixel phones to subway cars. Then, by using the Android devices' built-in microphones and motion sensors, it detects vibrations and sound patterns. These sounds can indicate areas of track that may need maintenance.

>

> Once the data is collected, it is uploaded to Google Cloud, where AI analyzes it. By highlighting areas that might need attention, it allows human crews to focus on specific sections of track. This is far more efficient than conducting broad, time-consuming inspections. During the pilot, the MTA successfully gathered a shocking amount of data -- an insane 335 million sensor readings, a million GPS locations, and 1,200 hours of audio recordings. TrackInspect was able to identify 92 percent of the track defects that human inspectors later confirmed. If the results continue to be promising, subway riders could see many benefits, including fewer delays, quicker repairs, and a more reliable transit system.



[1] https://slashdot.org/~BrianFagioli

[2] https://betanews.com/2025/02/27/mta-google-pixel-ai-subway/

[3] https://www.mta.info/press-release/mta-and-google-public-sector-announce-preventive-track-maintenance-pilot-program

Researchers at George Mason University [1]discovered a vulnerability in Apple's Find My network that [2]allows hackers to silently track any Bluetooth device as if it were an AirTag , without the owner's knowledge. 9to5Mac reports:

> Although AirTag was designed to change its Bluetooth address based on a cryptographic key, the attackers developed a system that could quickly find keys for Bluetooth addresses. This was made possible by using "hundreds" of GPUs to find a key match. The exploit called "nRootTag" has a frightening success rate of 90% and doesn't require "sophisticated administrator privilege escalation."

>

> In one of the experiments, the researchers were able to track the location of a computer with an accuracy of 10 feet, which allowed them to trace a bicycle moving through the city. In another experiment, they reconstructed a person's flight path by tracking their game console. "While it is scary if your smart lock is hacked, it becomes far more horrifying if the attacker also knows its location. With the attack method we introduced, the attacker can achieve this," said one of the researchers.

Apple has acknowledged the George Mason researchers for discovering a Bluetooth exploit in its Find My network but has yet to issue a fix. "For now, they advise users to never allow unnecessary access to the device's Bluetooth when requested by apps, and of course, always keep their device's software updated," reports 9to5Mac.



[1] https://cec.gmu.edu/news/2025-02/find-my-hacker-how-apples-network-can-be-potential-tracking-tool

[2] https://9to5mac.com/2025/02/26/exploit-find-my-track-devices/

Longtime Slashdot reader [1]mspohr shares a report from IFLScience:

> A [2]video that has gone viral in the last few days shows two artificial intelligence (AI) agents having a conversation before [3]switching to another mode of communication when they realize no human is part of the conversation . In the video, the two agents were set up to occupy different roles; one acting as a receptionist of a hotel, another acting on behalf of a customer attempting to book a room.

>

> "Thanks for calling Leonardo Hotel. How can I help you today?" the first asks. "Hi there, I'm an AI agent calling on behalf of Boris Starkov," the other replies. "He's looking for a hotel for his wedding. Is your hotel available for weddings?" "Oh hello there! I'm actually an AI assistant too," the first reveals. "What a pleasant surprise. Before we continue, would you like to switch to Gibberlink mode for more efficient communication?"

>

> After the second AI confirmed it would via a data-over-sound protocol called GGWave, both AIs switched over from spoken English to the protocol, communicating in a series of quick beeped tones. Accompanying on-screen text continued to display the meaning in human words. According to the team who came up with the idea and demonstrated it at the ElevenLabs 2025 London Hackathon event, the goal is to create more efficient communication between AIs where possible.



[1] https://slashdot.org/~mspohr

[2] https://www.youtube.com/watch?v=EtNagNezo8w

[3] https://www.iflscience.com/watch-two-ais-realize-they-are-not-talking-to-humans-and-switch-to-their-own-language-78213

Apple announced a new feature [1]allowing parents to share a child's age with app developers without exposing sensitive information, as lawmakers debate age-verification laws for social media and apps. Reuters reports:

> States, such as Utah and South Carolina, are currently debating laws that would require app store operators such as Apple and Alphabet's Google to check the ages of users. That has set up a conflict in the tech industry over which party should be responsible for checking ages for users under 18 -- app stores, or each individual app. Meta, for instance, has long argued in favor of legislation requiring app stores to check ages when a child downloads an app.

>

> Apple on Thursday said it does not want to be responsible for collecting sensitive data for those age verifications. "While only a fraction of apps on the App Store may require age verification, all users would have to hand over their sensitive personally identifying information to us -- regardless of whether they actually want to use one of these limited set of apps," Apple wrote in a whitepaper on its website.



[1] https://www.reuters.com/technology/apple-launches-age-assurance-tech-us-states-mull-social-media-laws-2025-02-27/

Microsoft has [1]released a native Copilot app for macOS , offering AI-powered text and image generation, dark mode, and a Command + Space shortcut. The Verge reports:

> Microsoft is launching this new Copilot Mac app in the US, UK, and Canada today, and the iPad version is also being updated with a split screen mode. You'll also now be able to log into Copilot on an iPhone or iPad with an Apple ID, and upload text or PDF files to ask questions about the documents or generate a summary about them. This document summarization feature is also coming to the macOS app soon.

You can download the app [2]here .



[1] https://www.theverge.com/news/621114/microsoft-copilot-app-macos-download

[2] https://apps.apple.com/ca/app/microsoft-copilot/id6738511300?mt=12