Latest news with #xAI
Yahoo
2 hours ago
- Business
- Yahoo
Meta wants to speed its race to ‘superintelligence'—but investors will still want their billions in ad revenue
Meta is doubling down on its so-called race to 'superintelligence,' reshuffling its AI organization once more as its new Meta Superintelligence Labs (MSL) group takes shape. But analysts say investors are keeping their eye on the prize Meta has always promised: improved products that increase engagement and, in turn, sell more ads. Superintelligent AI models, if they arrive, are just a means to that end. This time it's former Scale AI CEO Alexandr Wang—brought on by Mark Zuckerberg in June as chief AI officer—leading the reorganization. Wang, who now oversees a sprawling operation of thousands of engineers, scientists, and product managers, is looking to rein it in, reportedly resulting in some expected executive departures and at least one team shutdown. Wang was hired to help recruit a small, high-priced cadre of researchers—some reportedly offered compensation packages exceeding $100 million, typically spread out over several years—now perched at the pinnacle of Meta's AI effort. But that group is only the tip of the spear: The new restructuring folds the entire AI organization into MSL, with four new groups focused on research, training, products, and infrastructure, all part of a bid for speed. The quartet of group leaders will all report to Wang, including well-known investor and former GitHub CEO Nat Friedman, who will lead product and applied research, and former OpenAI researcher Shengjia Zhao, who will lead the research team as chief scientist. In a recent email to employees, which detailed the restructuring, Wang acknowledged that reorganizations can be disruptive but insisted the new structure would 'allow us to reach superintelligence with more velocity over the long term.' (Meta did not respond to Fortune's request to confirm the contents of the email, which were published by Business Insider.) Investors, meanwhile, seem to have mixed feelings: Meta's stock slid more than 2% on the news today, but climbed most of the way back by market close. The share-price slide also reflects broader market jitters, as overheated AI and Big Tech names come off recent highs, said Daniel Newman, CEO of research firm the Futurum Group. He said he expects a 'modest correction' but noted that Meta has 'had an incredible run' and recently 'delivered a great quarter once again.' Still, analysts are eyeing Zuckerberg's nine-figure paydays for top AI researchers and his repeated reorganizing, and watching for signs that Meta will close the gap in the AI race. 'Of course there is some concern,' Newman said, pointing out that numerous frontier models from OpenAI, xAI, and Google continue to improve, while Meta's open-source Llama models have 'seemingly stalled.' 'We think the team at Meta, after Zuckerberg's hiring spree, will need a period of acclimation before it finds the velocity to develop more competitive solutions,' he said. Feeding Meta's product machine That need for speed, however, is best understood as an extension of Meta's product machine rather than a bid to solve humanity's greatest challenges. While Meta has dabbled in moonshot AI through its FAIR research lab (cofounded by chief scientist Yann LeCun), rivals like OpenAI and Anthropic and spinoffs such as Thinking Machines Lab and Safe Superintelligence have made the pursuit of artificial general intelligence (AI generally defined to be as smart as humans) and superintelligence (AI far smarter than humans) their central mission, by contrast, has remained the same as it was before 'superintelligence' became a buzzword: improving the products that power engagement on its massively profitable social-media platforms, including Facebook, Instagram, and WhatsApp. The advertising on those platforms is the source of nearly all of Meta's revenue, which reached $46.6 billion in the most recent quarter. Zuckerberg underscored this focus last month with an Instagram Reel and blog post in which he said AI is rapidly advancing and that we're beginning to see 'glimpses of AI systems improving themselves.' Superintelligence is now 'in sight,' he added—but while rival AI companies talk about scientific or economic breakthroughs, his vision is aimed squarely at the individual: a personalized AI that helps you 'achieve your goals, create what you want to see in the world, be a better friend, and grow to become the person that you aspire to be.' That framing neatly aligns with what Meta has always built—consumer-facing experiences designed to keep people engaged (and sell more ads). To Zuckerberg, superintelligence also means powering the future of AI-infused personal devices, specifically augmented-reality glasses that can 'see what we see, hear what we hear, and interact with us throughout the day.' Newman said he continues to like Meta's prospects because the company 'isn't as dependent on the research end of its business, as it is using AI to continue to create higher daily active user numbers—and of course, the coinciding revenue continues to rise as well.' But Forrester's Mike Proulx countered that there is no doubt Meta is laser-focused on building 'the best and most powerful AI models, period,' he told Fortune. 'The race is on, and Meta is lagging against competitors. A concerted focus on superintelligence gives Meta a North Star to rally around both strategically and operationally.' Zuckerberg echoed that sentiment on Meta's most recent earnings call, stressing that AI is at the center of each of Meta's five focus areas. But Proulx pointed out that it was AI glasses—not the company's family of apps—that Zuckerberg highlighted on that call as 'the main way' superintelligence will enter people's daily lives. Overall, Proulx said he is not concerned with the seemingly constant upheaval in Meta's AI organization. 'This space is moving at breakneck speed. As with any emerging tech race, there's inevitably going to be a lot of pivoting. It comes with the territory,' he said. For all the lofty talk of superintelligence, however, Meta's AI reshuffling shows its bets are mostly still the same: personalized products that keep billions scrolling, ads flowing—and soon, AI-powered glasses perched on every face. How the company fares will be closely watched: 'The question now is whether the team is effectively enabled to deliver, or not,' said Proulx. This story was originally featured on 擷取數據時發生錯誤 登入存取你的投資組合 擷取數據時發生錯誤 擷取數據時發生錯誤 擷取數據時發生錯誤 擷取數據時發生錯誤
Engadget
3 hours ago
- Business
- Engadget
Oracle will reportedly power a giant data center with gas generators
Bloomberg has published a deep dive into operations at Oracle, chronicling the software giant's rise in cloud computing and current push into powering artificial intelligence projects. The publication reported that Oracle has promised to develop tens of billions of dollars in data centers, which have become a hot business. Notably, Oracle landed a deal to back operations at OpenAI, in a partnership that will give the AI company 4.5 gigawatts of computing power . According to Bloomberg , that's enough energy to power "millions of American homes." So far, Oracle has seemed willing to throw money at its AI data center projects, no matter how expensive or irresponsible the needs might be. For instance, a source said the company plans to spend more than $1 billion a year powering a single data center in Texas with gas generators rather than waiting for a utility connection to be built. When completed, this data center is expected to be one of the largest known sites, with computing power of 1.4 gigawatts. In addition to the huge monetary cost, such a project can also have negative human and environmental impact. The Elon Musk-owned xAI is under fire after a supercomputer for its artificial intelligence operations became a primary source of air pollution in Memphis thanks to methane-powered turbines. Other majors, including Google , Microsoft and Meta , have chosen to try nuclear power for their data center projects, which comes with its own potential complications and risks. Purely on the financial side, Oracle's decision to invest so much so quickly meant the company reported its first negative annual cash flow since 1990. Should the current rates of AI investment turn out to be a bubble, it could be very bad news to have many billions of dollars on the line.
Tom's Guide
6 hours ago
- Business
- Tom's Guide
Hundreds of thousands of Grok chatbot conversations are showing up in Google Search — here's what happened
If you've ever chatted with Grok, the Elon Musk-backed AI assistant from xAI, and hit the 'Share' button, your conversation might be searchable on Google. According to a recent Forbes report, more than 300,000 Grok conversations have been indexed by search engines, exposing user chats that were likely intended to be shared privately, not broadcast online. This comes just weeks after ChatGPT users discovered their shared chats were searchable on Google too. This discovery once again raises serious concerns about how AI platforms handle content sharing and user privacy. Built by Elon Musk's xAI, Grok includes a feature that lets users share conversations through a unique link. Once users share, those chats are no longer private, and are getting picked up by Google's web crawlers. From there, the chat are turning up in search results, meaning anyone can access and read them. In both the ChatGPT cases and this one, the issue comes down to how "shareable" URLs are structured and whether AI companies are doing enough to protect users who may not realize their content is publicly viewable. The implications go beyond just a few embarrassing screenshots. Some Grok chats could contain sensitive, personal or even private details. Many users likely had no idea their shared conversations were being published for the world to see until now. This privacy breach highlights a growing challenge in AI product design: how to balance transparency and social sharing with greater privacy protections. When users hit 'Share,' they should be clearly informed that their link will be publicly accessible, and AI companies need to use best practices like noindex tags or restricted access URLs to avoid surprises like this. To prevent these kinds of privacy slip-ups, AI platforms should: Get instant access to breaking news, the hottest reviews, great deals and helpful tips. This isn't the first controversy Grok has faced. the chatbot has come under fire in recent months for questionable responses and controversial content. But this latest issue hits at a core trust concern: how user data is managed and whether privacy is truly being prioritized. As more people turn to AI tools for personal, educational, wellness and emotional support, platforms like xAI will need to step up their safeguards or risk eroding the very trust they're trying to build. Although the damage is already done and deleting the chats won't make them completely disappear from searches, users can still take action. Start by safeguarding your chats by not sharing them until Grok has more secure privacy settings. Screenshots are another option if you must share your chats with others. As more people turn to AI tools for personal, educational, wellness and emotional support, platforms like xAI will need to step up their safeguards or risk eroding the very trust they're trying to Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
CNET
9 hours ago
- Business
- CNET
Hundreds of Thousands of User Chats with AI Chatbot Grok Are Now Public
You might assume that what's said between you and your AI chat stays between you and your AI chat? Unfortunately, that's not necessarily the case. According to a report by Forbes, Elon Musk's AI assistant Grok published more than 370,000 chats on the Grok website. Those URLs, which were not necessarily intended for public consumption by users, were then indexed by search engines and thus entered into the public sphere. And it wasn't just chats. Forbes also reported that uploaded documents such as photos, photos and spreadsheets were also published. Representatives for xAI, which makes Grok, didn't immediately respond to request for comment. It's the latest in a series of troubling reports that should spur chatbot users to be very careful about what they are sharing with AI assistants. Don't just mindlessly gloss over the Terms and Conditions -- be mindful of the privacy settings. Earlier this month, 404 Media reported on a researcher who discovered more than 130,000 chats with AI assistants Claud, Chat GPT and others were readable on When a Grok chat is finished, the user can hit a share button to create a unique URL, allowing the user to share the Grok conversation with others. These URLs were made available to search engines, allowing anyone to read them, the Forbes report said. There is no disclaimer that these chat URLs will be published for the open internet. But the Terms of Service outlined on the Grok website reads: "You grant, an irrevocable, perpetual, transferable, sublicensable, royalty-free, and worldwide right to xAI to use, copy, store, modify, distribute, reproduce, publish, display in public forums, list information regarding, make derivative works of, and aggregate your User Content and derivative works thereof for any purpose..."
Yahoo
10 hours ago
- Business
- Yahoo
Thousands of Grok chats are now searchable on Google
Hundreds of thousands of conversations that users had with Elon Musk's xAI chatbot Grok are easily accessible through Google Search, reports Forbes. Whenever a Grok user clicks the 'share' button on a conversation with the chatbot, it creates a unique URL that the user can use to share the conversation via email, text or on social media. According to Forbes, those URLs are being indexed by search engines like Google, Bing, and DuckDuckGo, which in turn lets anyone look up those conversations on the web. Users of Meta and OpenAI's chatbots were recently affected by a similar problem, and like those cases, the chats leaked by Grok give us a glimpse into users' less-than-respectable desires – questions about how to hack crypto wallets; dirty chats with an explicit AI persona; and asking for instructions on cooking meth. xAI's rules prohibit the use of its bot to 'promote critically harming human life' or developing 'bioweapons, chemical weapons, or weapons of mass destruction,' though that obviously hasn't stopped users from asking Grok for help with such things anyway. According to conversations made accessible by Google, Grok gave users instructions on making fentanyl, listed various suicide methods, handed out bomb construction tips, and even a detailed plan for the assassination of Elon Musk. xAI did not immediately respond to a request for comment. We've also asked when xAI began indexing Grok conversations. Late last month, ChatGPT users sounded the alarm that their chats were being indexed on Google, which OpenAI described as a 'short-lived experiment.' In a post Musk quote-tweeted with the words 'Grok ftw,' Grok explained that it had 'no such sharing feature' and 'prioritize[s] privacy.' Got a sensitive tip or confidential documents? We're reporting on the inner workings of the AI industry — from the companies shaping its future to the people impacted by their decisions. Reach out to Rebecca Bellan at and Maxwell Zeff at For secure communication, you can contact us via Signal at @rebeccabellan.491 and @mzeff.88.
