According to an SEC filing from NVIDIA, the US government now requires companies to obtain a license to export H20 integrated circuits and any other products that achieve the same performance benchmarks. The filing states that "the license requirement addresses the risk that the covered products may be used in, or diverted to, a supercomputer in China." Mainland China is not the only place targeted by this license; NVIDIA will also require permission to sell the H20 to the territories of Hong Kong and Macau as well as to nations with the D:5 designation as US Arms Embargo Countries.
The H20 chips are currently the most advanced chips that can be sold to select international markets under present laws and they are powerful enough to be used for artificial intelligence applications. NVIDIA has wanted the ability to retain Chinese customers for these products and last week, it seemed like the company may have gotten a reprieve on new restrictions. However, it appears that the new license requirement "will be in effect for the indefinite future."
NVIDIA said in the SEC filing that it now expects to report about $5.5 billion in charges related to "inventory, purchase commitments and related reserves" associated with the H20 circuits in the results for its current fiscal quarter.This article originally appeared on Engadget at https://www.engadget.com/big-tech/nvidia-says-the-us-has-put-export-restrictions-on-h20-ai-chips-224822930.html?src=rss
Two and a half centuries after Josiah Wedgwood perfected Jasperware, with its distinctive matte finish, the British ceramics brand is marking the milestone with a decidedly 21st-century approach. The company has launched Jasper 250 AI, a generative tool that enables anyone to get creative and play with an iconic style of pottery.The initiative echoes a 1930 international design competition when Wedgwood celebrated its founder's bicentennial, which crowned Danish artist Emmanuel Tjerne. Now, participants can share their AI-generated designs across social platforms using #jasper250, with the winning creation to be 3D-printed and acquired by the V&A Wedgwood Collection.Rather than asking people to passively admire tradition, Wedgwood is inviting them to actively collaborate with its legacy positioning AI as a bridge between today's consumers and the brand's heritage. By translating 250 years of stoneware artistry into an accessible digital format, Wedgwood demonstrates how even the most storied brands can allow consumers to mold tradition and make it their own.
Google announced today that it will no longer be using country code top level domains for searches. Instead, all search services will happen on the google.com URL and local results will be delivered automatically. For example, that means users in the UK will no longer see google.co.uk in their browser's address bar. Google URLs with those country-specific domain endings will now redirect to the main google.com address.
Google started using location information to automatically provide search results based on geography in 2017. With that change, it didn't matter whether you entered a query into a local country code URL or into google.com; you'd always see the results version for the place you were physically located. Today's announcement seems to take that initial action to its conclusion by sunsetting those ccTLDs.
"Its important to note that while this update will change what people see in their browser address bar, it wont affect the way Search works, nor will it change how we handle obligations under national laws," Google noted in its announcement.This article originally appeared on Engadget at https://www.engadget.com/big-tech/google-is-retiring-country-specific-domains-for-search-212157490.html?src=rss
In 2018, Mark Zuckerberg floated the idea of spinning out Instagram, one of the remedies the Federal Trade Commission (FTC) will likely seek in Meta's antitrust trial that began this week. CNBC reported on Tuesday that the comments from an email thread with executives came to light in Washington, DC.
"I'm beginning to wonder whether spinning Instagram out is the only structure that will accomplish a number of important goals," Zuckerberg wrote in the email. "As calls to break up the big tech companies grow, there is a non-trivial chance that we will be forced to spin out Instagram and perhaps WhatsApp in the next 5-10 years anyway." His estimate, made six years ago, ended up being spot-on.
"On the flip side, while most companies resist breakups, the corporate history is that most companies actually perform better after they've been split up," Zuckerberg added in the same email, according to The New York Times.
It's Zuckerberg's second day of testimony in the trial, which stemmed from a 2020 government lawsuit against Meta (then still known as Facebook). The FTC argues that the company's purchases of Instagram (for $1 billion in 2012) and WhatsApp (for $19 billion in 2014) hurt competition. If the trial goes the FTC's way, it will likely ask the judge to break up Meta by selling off one or both of the apps.
Last April, Meta moved to dismiss the case, but US District Judge James Boasberg allowed it to proceed with a narrower scope.
On the stand on Tuesday, the Meta CEO reportedly defended the company's Instagram purchase as the result of a standard cost-benefit examination. "We were doing a build-vs.-buy analysis," Zuckerberg said. "I thought that Instagram was better at that [than Facebook's Camera app], so I thought it was better to buy them."
"Building a new app is hard," he said when asked on the stand why he was intent on buying Instagram. "Weve probably tried building dozens of apps over the history of the company, and the majority of them dont go anywhere."
Other details that came to light on Tuesday include a 2013 email in which Zuckerberg told executives to block Asian competitors Kakao and WeChat from advertising on Facebook. "Those companies are trying to build social networks and replace us," he wrote. "The revenue is immaterial to us compared to any risk."This article originally appeared on Engadget at https://www.engadget.com/big-tech/mark-zuckerberg-predicted-metas-antitrust-trial-in-a-2018-email-205719635.html?src=rss
Kia has teased that it will have a trio of announcements at the 2025 New York International Auto Show. The car company's presentation is scheduled for 10AM ET on April 16. Luckily for those gearheads who want to know what's next for the brand, Kia will have a livestream of its showcase. The reveal will be hosted live on YouTube, and it's also embedded below.
The presentation will feature three debuts, according to the company, including two that are fully electric. Kia unveiled its EV4 hatchback sedan in February, but didn't share info about North American manufacturing at that time. That's a likely announcement during this week's New York event.
While the stage presentation will get livestreamed, we don't yet have tech to remotely test the Kia EV9 and EV6 models that will be at the live show. Maybe one dayThis article originally appeared on Engadget at https://www.engadget.com/transportation/how-to-watch-kias-new-york-international-auto-show-presentation-on-april-16-195259387.html?src=rss
Whether in Ancient Greek parodies, Shakespeare's subtle jabs at the haughty upper class or Jon Stewart's takedowns of the latest Washington absurdities, humor can be a razor-edged weapon that shapes public opinion. Last weekend, we were reminded again of comedy's ability to strike at the rich and powerful when someone hacked Silicon Valley crosswalks to play synthetic voice impersonations of Mark Zuckerberg and Elon Musk.
"Hi, this is Elon Musk," the voice impersonating the Trump ally and federal government dismantler said. "Welcome to Palo Alto, the home of Tesla Engineering. You know, they say money can't buy happiness. And yeah, okay I guess that's true. God knows I've tried. But it can buy a Cybertruck, and that's pretty sick, right? Right?? Fuck, I'm so alone."
"Hi, this is Mark Zuckerberg, but real ones call me the Zuck," the synthetic voice of Meta's CEO said over one crosswalk speaker. "It's normal to feel uncomfortable or even violated as we forcefully insert AI into every facet of your conscious experience. And I just want to assure you, you don't need to worry because there's absolutely nothing you can do to stop it. Anyway, see ya."
The videos have accumulated over 100,000 likes on TikTok and nearly 400,000 views on X (ironically). Palo Alto Online reports (via TechCrunch) that they could be heard in downtown intersections in Redwood City, Menlo Park and Palo Alto but had been removed beginning on Saturday.
"Hi, this is Elon Musk," another crosswalk speaker emitted. "And I'd like to personally welcome you to Palo Alto. People keep saying cancer is bad, but have you tried being a cancer? It's fucking awesome."
"Hey, it's Zuck here," another voice said. "I just wanted to tell you how very proud I am of everything we've been building together. From undermining democracy to cooking our grandparents' brains with AI slop to making the world less safe for trans people, nobody does it better than us. And I think that's pretty neat. Zuck out!"
Finally, a Palo Alto speaker conducted armchair psychology on the Tesla CEO. "Hi, I'm Elon. Can we be friends? Will you be my friend? I'll give you a Cybertruck, I promise. Okay, look, you don't know the level of depravity I would stoop to just for a crumb of approval. I mean, let's be real. It's not like I had any moral convictions to begin with, right?"
This article originally appeared on Engadget at https://www.engadget.com/cybersecurity/someone-hacked-silicon-valley-crosswalks-to-spoof-mark-zuckerberg-and-elon-musk-191822645.html?src=rss
Anthropic announced that its Claude AI can integrate with Google Workspace. This tie-in allows the AI assistant to access any information in Gmail, Google Documents and Google Calendar. Enterprise-level customers even get a special cataloguing option for Documents that aims to offer even better speed and accuracy when retrieving information.
This update could make Claude more helpful when it comes to using the chatbot for scheduling or accessing information within the Google ecosystem. The blog post with the announcement specified that the Enterprise option comes with special security controls for confidentiality, but doesn't detail if or how other users might be able to keep Claude from accessing sensitive information that might be stored in an email or document. Google Workspace integration is available in beta now for all paid Anthropic customers.
Anthropic is also adding a Research feature. Queries in this mode are intended to offer thorough answers to queries that "explores different angles of your question automatically and works through open questions systematically." Claude's responses in Research will include citations for fact-checking. Anthropic says this feature can be combined with the Google integration for analyzing information stored across multiple different locations, such as notes about complicated work or school projects. Research is available in the US, Japan and Brazil as an early beta for users on the Max, Team and Enterprise plans.This article originally appeared on Engadget at https://www.engadget.com/ai/anthropics-claude-can-now-read-your-emails-181511019.html?src=rss
It looks like OpenAI is building its own X-like social media network, according to a report by The Verge. We dont have many specifics, but we do know theres an internal prototype that adds a social feed to ChatGPTs image generation tool.
It remains unclear if OpenAI will launch this social network as a standalone app or if it will be integrated within the ChatGPT app, which is what the prototype indicates. The report does suggest that OpenAI CEO Sam Altman has been asking for feedback about the social network from people outside of the company.
One potential reason for this step is that the app would allow OpenAI to gather real-time data from users to train its AI models. Its also a potentially diversifying move for the company.
Although its uncertain if this project will ever see the light of day, one thing is probable itll likely get under the skin of Elon Musk, who owns X. Altman and OpenAI have been at odds with Musk for some time. The company rejected his offer to buy OpenAI, instead offering to "buy Twitter" for just $9.7 billion dollars. Its worth noting that the offer is four times lower than what Musk paid for the platform and that Altman purposefully used the word Twitter and not X.
no thank you but we will buy twitter for $9.74 billion if you want Sam Altman (@sama) February 10, 2025
OpenAI has also recently countersued Musk, accusing the billionaire of staging press attacks and malicious campaigns against the company on X. The suit also accuses Musk of making "harassing legal claims," referring to his initial suit, and a "sham bid for OpenAI's assets."
The company also published an email from Musks time at OpenAI in which he suggested merging with Tesla. Altman and his crew have stated that Musk was present during talks about OpenAI going for-profit and that he suggested he be given majority equity, control of the board and the CEO position. Musk left OpenAI in 2018 and founded xAI, which somehow just bought X in some kind of shady ouroboros of corporate tomfoolery.This article originally appeared on Engadget at https://www.engadget.com/ai/openai-is-apparently-making-a-social-network-175842370.html?src=rss
With America's long-term (and even short-term) economic prospects shifting on Eric Cartman-like whims, some internet providers have begun offering multi-year price guarantees to provide a little stability amid the chaos. A couple of weeks after Verizon introduced a three-year price lock, Comcast has followed suit with a five-year pricing guarantee for new Xfinity residential customers.
On the bright side, the deal isn't linked to an annual contract requirement. But, as you might expect from a massive broadband company, there's plenty of fine print.
First, the offer only applies to 400 Mbps home internet plans with one year of Xfinity Mobile Unlimited (one line) a package that starts at $55 monthly (before taxes and fees). Then, if you want unlimited data, you'll need to use an Xfinity Gateway, which requires a monthly rental payment, or pay an extra $30 monthly to enjoy it with your own modem.
The locked-in price doesn't cover installation, taxes, fees or other charges, which Comcast notes are subject to change during and after the promo. So, at least in theory, the company has some wiggle room to raise the price without technically raising it. Finally, if you don't enroll in and maintain paperless billing and automatic payments, you'll pay an extra $10 each month (or an extra $8 if you pay by credit or debit card).
After reading those two paragraphs of fine print, you may find it amusing that Comcast's press release frames the price guarantee as being about simplicity. "Our new price guarantee removes complexity," Comcast COO Steve Croney wrote. Still, with Americans increasingly being forced to check their 401(k)s and stock portfolios like gamblers check a high-stakes sports score, any small oasis of financial stability is at least something.
Comcast's offer ends on June 23.This article originally appeared on Engadget at https://www.engadget.com/big-tech/comcast-rolls-out-a-five-year-price-guarantee-with-plenty-of-fine-print-173421738.html?src=rss
Following a limited, four-month preview, Google is making its Veo 2 video generation model available to more people. The company has begun rolling out the tool to Gemini Advanced users, who can use it to create 720p clips that are up to eight seconds long. "The more detailed your description, the more control you have over the final video," Google says of the model.
For the time being, Gemini will only generate footage in 16:9 landscape orientation, meaning it wont be great for creating social video. That said, Google notes you can upload videos directly to TikTok or YouTube through Gemini's share sheet, and any footage you create through the chatbot will automatically be formatted as a MP4 file. Relative to the models iteration, Google says Veo 2 has a better understanding of real-world physics and human motion.
If you want to start using Gemini to generate video clips, you'll need Google's $20 per month One AI Premium Plan. Once the company completes the roll out of Veo 2, video generation will be available in every language you can use Gemini on both on the web and mobile. Separately, Google is bringing Veo 2 generation to Whisk. For the uninitiated, Whisk is an experimental AI application the company offers through its Google Labs beta platform. It allows you to generate images from both text and image prompts, and with Veo 2 now baked in, you can use Whisk to animate what you create through the app.This article originally appeared on Engadget at https://www.engadget.com/ai/google-brings-veo-2-video-generation-to-gemini-advanced-170014422.html?src=rss
