Apple has long avoided directly purchasing Nvidia's chips and is now developing its own AI server chip with Broadcom, aiming for production by 2026, The Information reported Tuesday, shedding broader light on why the two companies don't get along so well.

The relationship deteriorated after a 2001 meeting where Steve Jobs accused Nvidia of copying technology from Pixar, which he then controlled. Relations worsened in 2008 when Nvidia's faulty graphics chips forced Apple to extend MacBook warranties without full compensation.

Rather than buying Nvidia's dominant AI processors directly like its tech peers, the Information reports, Apple rents them through cloud providers while also using Google's custom chips for training large AI models. The company's new chip project, code-named Baltra, marks its most ambitious effort yet to reduce reliance on external AI processors, despite being one of the largest indirect users of Nvidia chips through cloud services.

Contractors working to improve Google's Gemini AI are comparing its answers against outputs produced by Anthropic's competitor model Claude, TechCrunch reported Tuesday, citing internal correspondence. From the report: Google would not say, when reached by TechCrunch for comment, if it had obtained permission for its use of Claude in testing against Gemini.

As tech companies race to build better AI models, the performance of these models are often evaluated against competitors, typically by running their own models through industry benchmarks rather than having contractors painstakingly evaluate their competitors' AI responses. The contractors working on Gemini tasked with rating the accuracy of the model's outputs must score each response that they see according to multiple criteria, like truthfulness and verbosity. The contractors are given up to 30 minutes per prompt to determine whose answer is better, Gemini's or Claude's, according to the correspondence seen by TechCrunch.

An anonymous reader quotes a report from Gizmodo: Once an icon of the 20th century seen as obsolete in the 21st, Encyclopedia Britannica -- now known as just Britannica -- is all in on artificial intelligence, and may soon go public at a valuation of nearly $1 billion, according to the New York Times.

Until 2012 when printing ended, the company's books served as the oldest continuously published, English-language encyclopedias in the world, essentially collecting all the world's knowledge in one place before Google or Wikipedia were a thing. That has helped Britannica pivot into the AI age, where models benefit from access to high-quality, vetted information. More general-purpose models like ChatGPT suffer from hallucinations because they have hoovered up the entire internet, including all the junk and misinformation.

While it still offers an online edition of its encyclopedia, as well as the Merriam-Webster dictionary, Britannica's biggest business today is selling online education software to schools and libraries, the software it hopes to supercharge with AI. That could mean using AI to customize learning plans for individual students. The idea is that students will enjoy learning more when software can help them understand the gaps in their understanding of a topic and stay on it longer. Another education tech company, Brainly, recently announced that answers from its chatbot will link to the exact learning materials (i.e. textbooks) they reference.

Britannica's CEO Jorge Cauz also told the Times about the company's Britannica AI chatbot, which allows users to ask questions about its vast database of encyclopedic knowledge that it collected over two centuries from vetted academics and editors. The company similarly offers chatbot software for customer service use cases. Britannica told the Times it is expecting revenue to double from two years ago, to $100 million.

An anonymous reader shares a report: The Department of Justice's list of solutions for fixing Google's illegal antitrust behavior and restoring competition in the search engine market started with forcing the company to sell Chrome, and late Friday night, Google responded with a list of its own.

Instead of breaking off Chrome, Android, or Google Play as the DOJ's filing considers, Google's proposed fixes aim at the payments it makes to companies like Apple and Mozilla for exclusive, prioritized placement of its services, its licensing deals with companies that make Android phones, and contracts with wireless carriers. They don't address a DOJ suggestion about possibly forcing Google to share its valuable search data with other companies to help their products catch up.

Meta plans to add displays to its Ray-Ban smart glasses as soon as next year, Financial Times reports, as the US tech giant accelerates its plans to build lightweight headsets that can usurp the smartphone as consumers' main computing device. Financial Times: The $1.5tn social media group is planning to add a screen inside the $300 sunglasses it makes and sells in partnership with eyewear group EssilorLuxottica, according to people familiar with the plans. The updated Ray-Bans could be released as early as the second half of 2025, the people said. The small display would be likely to be used to show notifications or responses from Meta's virtual assistant.

The move comes as Meta pushes further into wearable devices and what chief executive Mark Zuckerberg hopes will be the next computing platform, as rivals such as Apple, Google and Snap also race to develop their own similar products.

Newsweek supplies some context After his arrest, merch — including T-shirts featuring Mangione's booking photos and others taken from his social media accounts — began popping up for sale on several sites. Websites, including Amazon, eBay and Etsy, have moved to take down products that glorify violence or the suspect. An eBay spokesperson told Newsweek that "items that glorify or incite violence, including those that celebrate the recent murder of UHC CEO Brian Thompson, are prohibited."
Inc. magazine adds: Separately, GoFundMe has shuttered several fundraising campaigns created for Mangione. The fundraising site's terms and conditions are pretty clear on the matter, NBC News reports, with a company spokesperson explaining they prohibit "fundraisers for the legal defense of violent crimes."
But one incident was different, according to a post from the law school of the University of British Columbia: To provide a quick summary, Rachel Kenaston, an artist selling merch on TeePublic received an e-mail from the platform regarding intellectual property claim by UnitedHealth Group Inc and decided to remove Kenaston's design from the merch store. Obviously, it is important to point out that it isn't quite clear who is filing those DMCA claims. While TeePublic, in the email, claimed that they have no say in the matter, [an article from 404 Media] goes on to explain that TeePublic has the right to refuse DMCA claims, but often choose not to in order to avoid headache. The design had nothing to do with UnitedHealthcare-it seems to be a picture of the Mangione in a heart frame. Meaning, whether it was UnitedHealthcare or not, the claim shouldn't hold any weight.

Consensus seems to be mostly leaning towards speculation that it is unlikely to be UnitedHealthcare actually filing those DMCA claims, but rather potential competitors... Regardless of whether or not it really was UnitedHealthcare that filed DMCA claims, I think the important point here is that the merch actually did get taken down. In fact, this would be more problematic if it was from a competitor using DMCA as a form of removing competition, because, then it really has nothing to do with intellectual property. I would assume that this happens quite frequently. Especially for YouTubers, it seems that copyright strikes are more than a mere pesky occurrence, but for many, something that affects livelihood...

The difficult part, as always, is finding the balance between protecting the rights of the copyright holders and ensuring that the mechanisms doesn't get abused.
The artist told Gizmodo she was filing a counterclaim to the copyright notice, adding that instead of a DMCA, "I honestly expected the design to be pulled for condoning violence or something..."

Gizmodo published the image — a watercolored rendition of a hostel surveillance-camera photo released by police — adding "UnitedHealth Group didn't respond to questions emailed on Monday [December 16] about how the company could possibly claim a copyright violation had occurred." And while Gizmodo promised they'd update the post if UnitedHealth responded — there has been no update since...

404 Media adds that the watercolor "is not the only United Healthcare or Luigi Mangione-themed artwork on the internet that has been hit with bogus DMCA takedowns in recent days. Several platforms publish the DMCA takedown requests they get on the Lumen Database, which is a repository of DMCA takedowns." On December 7, someone named Samantha Montoya filed a DMCA takedown with Google that targeted eight websites selling "Deny, Defend, Depose" merch that uses elements of the United Healthcare logo... Medium, one of the targeted websites, has deleted the page that the merch was hosted on...

Over the weekend, a lawyer demanded that independent journalist Marisa Kabas take down an image of Luigi Mangione and his family that she posted to Bluesky, which was originally posted on the campaign website of Maryland assemblymember Nino Mangione. The lawyer, Desiree Moore, said she was "acting on behalf of our client, the Doe Family," and claimed that "the use of this photograph is not authorized by the copyright owner and is not otherwise permitted by law..." In a follow-up email to Kabas, Moore said "the owner of the photograph has not authorized anyone to publish, disseminate, or otherwise use the photograph for any purpose, and the photograph has been removed from various digital platforms as a result," which suggests that other websites have also been threatened with takedown requests. Moore also said that her "client seeks to remain anonymous" and that "the photograph is hardly newsworthy."
404 Media believes the takedown request "shows that the Mangione family or someone associated with it is using the prospect of a copyright lawsuit to threaten journalists for reporting on one of the most important stories of the year..."

UPDATE: Long-time Slashdot reader destinyland notes there's an interesting precedent from 2007: [D]eep within the DMCA law is a counter-provision — 512(f), which states that misrepresenting yourself as a copyright owner has consequences. Any damage caused by harmful misrepresentation must be reimbursed. In 2004 the Electronic Frontier Foundation won a six-figure award from Diebold Election Systems, who had claimed a "copyright" on embarrassing internal memos which were published online.

Monday long-time Slashdot reader denelson83 wrote: The crowdsourced, widely-used map database OpenStreetMap has had a hardware failure at its upstream ISP in Amsterdam and has been put into a protective read-only mode to avoid loss or corruption of data. .
The outage had started Sunday December 15 at 4:00AM (GMT/UTC), but by Tuesday they'd posted a final update: Our new ISP is up and running and we have started migrating our servers across to it. If all goes smoothly we hope to have all services back up and running this evening...

We have dual redundant links via separate physical hardware from our side to our Tier 1 ISP. We unexpectedly discovered their equipment is a single point failure. Their extended outage is an extreme disappointment to us.

We are an extremely small team. The OSMF budget is tiny and we could definitely use more help. Real world experience... Ironically we signed a contract with a new ISP in the last few days. Install is on-going (fibre runs, modules & patching) and we expect to run old and new side-by-side for 6 months. Significantly better resilience (redundant ISP side equipment, VRRP both ways, multiple upstream peers... 2x diverse 10G fibre links).
OpenStreetMap celebrated its 20th anniversary in August, with a TechCrunch profile reminding readers the site gives developers "geographic data and maps so they can rely a little less on the proprietary incumbents in the space," reports TechCrunch, adding "Yes, that mostly means Google."

OpenStreetMap starts with "publicly available and donated aerial imagery and maps, sourced from governments and private organizations such as Microsoft" — then makes them better: Today, OpenStreetMap claims more than 10 million contributors who map out and fine-tune everything from streets and buildings, to rivers, canyons and everything else that constitutes our built and natural environments... Contributors can manually add and edit data through OpenStreetMap's editing tools, and they can even venture out into the wild and map a whole new area by themselves using GPS, which is useful if a new street crops up, for example...

OpenStreetMap's Open Database License allows any third-party to use its data with the appropriate attribution (though this attribution doesn't always happen). This includes big-name corporations such as Apple and VC-backed unicorns like MapBox, through a who's who of tech companies, including Uber and Strava... More recently, the Overture Maps Foundation — an initiative backed by Microsoft, Amazon, Meta and TomTom — has leaned heavily on OpenStreetMap data as part of its own efforts to build a viable alternative to Google's walled mapping garden.
The article notes that OpenStreetMap is now overseen by the U.K.-based nonprofit OpenStreetMap Foundation (supported mainly by donations and memberships), with just one employee — a system engineer — "and a handful of contractors who provide administrative and accounting support."

In August its original founder Steve Coast, returned to the site for a special blog post on its 20th anniversary: OpenStreetMap has grown exponentially or quadratically over the last twenty years depending on the metric you're interested in... The story isn't so much about the data and technology, and it never was. It's the people... OpenStreetMap managed to map the world and give the data away for free for almost no money at all. It managed to sidestep almost all the problems that Wikipedia has by virtue of only representing facts not opinions. The project itself is remarkable. And it's wonderful that so many are in love with it.
"Two decades ago, I knew that a wiki map of the world would work," Coast writes. "It seemed obvious in light of the success of Wikipedia and Linux..."

An anonymous reader shared this report from TechCrunch: Microsoft-owned GitHub announced on Wednesday a free version of its popular Copilot code completion/AI pair programming tool, which will also now ship by default with Microsoft's popular VS Code editor. Until now, most developers had to pay a monthly fee, starting at $10 per month, with only verified students, teachers, and open source maintainers getting free access...

There are some limitations to the free version, which is geared toward occasional users, not major work on a big project. Developers on the free plan will get access to 2,000 code completions per month, for example, and as a GitHub spokesperson told me, each Copilot code suggestion will count against this limit — not just accepted suggestions. And while GitHub recently added the ability to switch between different foundation models, users on the free plan are limited to Anthropic's Claude 3.5 Sonnet and OpenAI's GPT-4o. (The paid plans also include Google's Gemini 1.5 Pro and OpenAI's o1-preview and -mini.) For Copilot Chat, the number of chat messages is limited to 50, but otherwise, there aren't any major limitations to the free service. Developers still get access to all Copilot Extensions and skills.

The free Copilot SKU will work in a number of editors, including VS Code, Visual Studio, and JetBrains, as well as on GitHub.com.
GitHub's announcement ends with the words "Happy coding!" and calls the service "GitHub Copilot Free." But TechCrunch points out there's already competition from services like Amazon Q Developer, as well as from companies like Tabnine and Qodo (previously known as Codium) — and they typically offer a free tier. But in addition, "With Copilot Free, we are returning to our freemium roots," GitHub CEO Thomas Dohmke told TechCrunch, as well as "laying the groundwork for something far greater: AI represents our best path to enabling a GitHub with one billion developers.

"There should be no barrier to entry for experiencing the joy of creating software. Now six years after being acquired by Microsoft, it indeed appears GitHub is still GitHub — and we are doing our thing."

Or, as GitHub CEO Satya Nadella said in a video posted on LinkedIn, "The joy of coding is back! And we are looking forward to bringing the same experience to so many more people around the world."

The rise of AI has rendered traditional CAPTCHA tests increasingly ineffective, as bots can now "[solve] these puzzles in milliseconds using artificial intelligence (AI)," reports The Conversation. "How ironic. The tools designed to prove we're human are now obstructing us more than the machines they're supposed to be keeping at bay." The report warns that the imminent arrival of AI agents -- software programs designed to autonomously interact with websites on our behalf -- will further complicate matters. From the report: Developers are continually coming up with new ways to verify humans. Some systems, like Google's ReCaptcha v3 (introduced in 2018), don't ask you to solve puzzles anymore. Instead, they watch how you interact with a website. Do you move your cursor naturally? Do you type like a person? Humans have subtle, imperfect behaviors that bots still struggle to mimic. Not everyone likes ReCaptcha v3 because it raises privacy issues -- plus the web company needs to assess user scores to determine who is a bot, and the bots can beat the system anyway. There are alternatives that use similar logic, such as "slider" puzzles that ask users to move jigsaw pieces around, but these too can be overcome.

Some websites are now turning to biometrics to verify humans, such as fingerprint scans or voice recognition, while face ID is also a possibility. Biometrics are harder for bots to fake, but they come with their own problems -- privacy concerns, expensive tech and limited access for some users, say because they can't afford the relevant smartphone or can't speak because of a disability. The imminent arrival of AI agents will add another layer of complexity. It will mean we increasingly want bots to visit sites and do things on our behalf, so web companies will need to start distinguishing between "good" bots and "bad" bots. This area still needs a lot more consideration, but digital authentication certificates are proposed as one possible solution.

In sum, Captcha is no longer the simple, reliable tool it once was. AI has forced us to rethink how we verify people online, and it's only going to get more challenging as these systems get smarter. Whatever becomes the next technological standard, it's going to have to be easy to use for humans, but one step ahead of the bad actors. So the next time you find yourself clicking on blurry traffic lights and getting infuriated, remember you're part of a bigger fight. The future of proving humanity is still being written, and the bots won't be giving up any time soon.

An anonymous reader quotes a report from TechCrunch: The newest online-only school greenlighted (PDF) by the Arizona State Board for Charter Schools comes with a twist: The academic curriculum will be taught entirely by AI. Charter schools -- independently operated but publicly funded -- typically get greater autonomy compared to traditional public schools when it comes to how subjects are taught. But Unbound Academy's application, which proposes an "AI-driven adaptive learning technology" that "condenses academic instruction into a two-hour window," is a first for the model. (Unbound's founders have been running a similar program at a "high-end private school" in Texas, which appears to be in-person.)

Unbound's approach leans on edtech platforms like IXL and Khan Academy, and students engage with "interactive, AI-powered platforms that continuously adjust to their individual learning pace and style." There will be humans, just fewer of them, and maybe not actual accredited teachers: It will adopt a "human-in-the-loop" approach with "skilled guides" monitoring progress who can provide "targeted interventions" and coaching for each student. Academic instruction is whittled down to just two hours. The remainder of the students' day will include "life-skills workshops" covering areas such as critical thinking, creative problem-solving, financial literacy, public speaking, goal setting, and entrepreneurship. The online-only school targets students from fourth to eighth grades.

OpenAI has unveiled a new AI model that it says takes longer to solve problems but gets better results, following Google's similar announcement a day earlier. The model, called o3, replaces o1 from September and spends extra time working through questions that need step-by-step reasoning.

It scores three times higher than o1 on ARC-AGI, a test measuring how well AI handles complex math and logic problems it hasn't seen before. "This is the beginning of the next phase of AI," CEO Sam Altman said during a livestream Friday.

The Microsoft-backed startup is keeping o3 under wraps for now but plans to let outside researchers test it.

Google has reduced its senior management positions by 10% as part of an ongoing efficiency initiative, CEO Sundar Pichai announced during a company-wide meeting earlier this week.

The restructuring affected managers, directors, and vice presidents, with some roles eliminated and others converted to non-management positions, a Google spokesperson told BusinessInsider. The move follows Google's January 2023 layoff of 12,000 employees and Pichai's September 2022 goal to improve company efficiency by 20%.

Home Assistant (not to be confused with the Google Assistant on Google Home) has launched the Voice Preview Edition (Voice PE), its first dedicated voice assistant hardware for $59. The device offers a privacy-focused, locally controlled solution that supports over 50 languages and integrates seamlessly with the open-source smart home platform. As The Verge notes, Voice PE supports the wake words "Hey Jarvis" right out of the box. From the report: The Voice PE is a small white box, about the size of your palm, with dual microphones and an audio processor. An internal speaker lets you hear the assistant, but you can also connect a speaker to it via a 3.5 mm headphone jack for better-quality media playback. A colored LED ring on top of the Voice PE indicates when the assistant is listening. It surrounds a rotary dial and a physical button, which is used for setup and to talk to the voice assistant without using the wake word. The button can also be customized to do whatever you want (because this is Home Assistant). A physical mute switch is on the side, and the device is powered by USB-C (charger and cable not included). There's also a Grove port where you can add sensors and other accessories.

For those who don't like the idea of always-listening microphones in their home from companies such as Amazon and Google, but who still want the convenience of controlling their home with their voice, the potential here is huge. But it may be a while until Voice PE is ready to replace your Echo or Nest smart speaker. [...] if you want more features, Voice PE can connect to supported AI models, such as ChatGPT or Gemini, to fully replace Assist or use it as a fallback for commands it doesn't understand. But for many smart home users, there will be plenty of value in a simple, inexpensive device that lets you turn your lights on and off, start a timer, and execute other useful commands with your voice without relying on an internet connection.

An anonymous reader quotes a report from TechCrunch: Google has released what it's calling a new "reasoning" AI model -- but it's in the experimental stages, and from our brief testing, there's certainly room for improvement. The new model, called Gemini 2.0 Flash Thinking Experimental (a mouthful, to be sure), is available in AI Studio, Google's AI prototyping platform. A model card describes it as "best for multimodal understanding, reasoning, and coding," with the ability to "reason over the most complex problems" in fields such as programming, math, and physics. [...]

Built on Google's recently announced Gemini 2.0 Flash model, Gemini 2.0 Flash Thinking Experimental appears to be similar in design to OpenAI's o1 and other so-called reasoning models. Unlike most AI, reasoning models effectively fact-check themselves, which helps them avoid some of the pitfalls that normally trip up AI models. As a drawback, reasoning models often take longer -- usually seconds to minutes longer -- to arrive at solutions. Given a prompt, Gemini 2.0 Flash Thinking Experimental pauses before responding, considering a number of related prompts and "explaining" its reasoning along the way. After a while, the model summarizes what it considers to be the most accurate answer.

Writers, publishers, musicians, photographers, movie producers and newspapers have rejected the Labour government's plan to create a copyright exemption to help AI companies train their algorithms. From a report: In a joint statement, bodies representing thousands of creatives dismissed the proposal made by ministers on Tuesday that would allow companies such as Open AI, Google and Meta to train their AI systems on published works unless their owners actively opt out.

The Creative Rights in AI Coalition (Crac) said existing copyright laws must be respected and enforced rather than degraded. The coalition includes the British Phonographic Industry, the Independent Society of Musicians, the Motion Picture Association and the Society of Authors as well as Mumsnet, the Guardian, Financial Times, Telegraph, Getty Images, the Daily Mail Group and Newsquest.

Their intervention comes a day after the technology and culture minister Chris Bryant told parliament the proposed system, subject to a 10-week consultation, would "improve access to content by AI developers, whilst allowing rights holders to control how their content is used for AI training."

Spanish police have uncovered a major clue in the year-long investigation of a missing Cuban man, JLPO, after Google Street View images showed a man loading a body-shaped package into a car and pushing a wheelbarrow with a large white package. These images led to the discovery of the victim's dismembered remains in a cemetery and the arrest of two suspects, including the victim's wife and a bar worker. The Independent reports: Spanish police have said the pictures are a "decisive" clue in case, with detectives reportedly launching a murder investigation and arresting two people in connection with the man's death. According to El Pais, police are still investigating the case -- and it appears neither have yet appeared charged before a court.

Microsoft bought twice as many of Nvidia's flagship chips as any of its largest rivals in the US and China this year, as OpenAI's biggest investor accelerated its investment in artificial intelligence infrastructure. From a report: Analysts at Omdia, a technology consultancy, estimate that Microsoft bought 485,000 of Nvidia's "Hopper" chips this year. That put Microsoft far ahead of Nvidia's next biggest US customer Meta, which bought 224,000 Hopper chips, as well as its cloud computing rivals Amazon and Google.

With demand outstripping supply of Nvidia's most advanced graphics processing units for much of the past two years, Microsoft's chip hoard has given it an edge in the race to build the next generation of AI systems. This year, Big Tech companies have spent tens of billions of dollars on data centres running Nvidia's latest chips, which have become the hottest commodity in Silicon Valley since the debut of ChatGPT two years ago kick-started an unprecedented surge of investment in AI.

An anonymous reader quotes a report from the New York Times: TikTokasked the Supreme Court on Monday to temporarily block a law that would effectively ban it in the United States in a matter of weeks. Saying that the law violates both its First Amendment rights and those of its 170 million American users, TikTok, which is controlled by a Chinese parent company, urged the justices to maintain the status quo while they decide whether to hear an appeal. "Congress's unprecedented attempt to single out applicants and bar them from operating one of the most significant speech platforms in this nation presents grave constitutional problems that this court likely will not allow to stand," lawyers for TikTok wrote in their emergency application.

President Biden signed the law this spring after it was enacted with wide bipartisan support. Lawmakers said the app's ownership represented a risk because the Chinese government's oversight of private companies would allow it to retrieve sensitive information about Americans or to spread propaganda, though they have not publicly shared evidence that this has occurred. They have also noted that American platforms like Facebook and YouTube are banned in China, and that TikTok itself is not allowed in the country.

YouTube is introducing an optional feature allowing creators to let third-party companies use their videos to train AI models, with the default setting being opt-out. The Verge reports: "We see this as an important first step in supporting creators and helping them realize new value for their YouTube content in the AI era," a TeamYouTube staffer named Rob says in a support post. "As we gather feedback, we'll continue to explore features that facilitate new forms of collaboration between creators and third-party companies, including options for authorized methods to access content."

YouTube will be rolling out the setting in YouTube Studio "over the next few days," and unauthorized scraping "remains prohibited," Rob writes. Another support page says that you'll be able to pick and choose from a list of third-party companies that can train on your videos or you can simply allow all third-party companies to train on them.

Cloudflare's 2024 internet traffic report highlights a 17.2% global increase in traffic, with Google maintaining its position as the most visited service and the U.S. responsible for 34.6% of bot traffic. The Register reports: One surprise (or perhaps not) is that IPv6 traffic is actually down as a percentage of the packets that passed through Cloudflare's network. It says that 28.5 percent of global traffic was IPv6 during 2024, whereas last year's report put this figure at 33.75 percent. The company also reveals that a fifth of all TCP connections (20.7 percent) are unexpectedly terminated before any useful data can be exchanged. Causes of this could vary from DoS attacks, quirky client behavior, or a network interrupting a connection to filter content.

Coudflare says about half of these incidents were connections closed "Post SYN" -- after its server has received a client's SYN packet, but before a subsequent acknowledgement (ACK) or any useful data. These can be attributed to DoS attacks or internet scanning, while Post-ACK or Post-PSH anomalies are more often associated with connection tampering activity such as filtering, especially if they occur at high rates in specific networks. Mobile device traffic accounted for about 41.3 percent of the total, which is roughly the same as last year. This is largely split between the Apple and Android ecosystems, with iOS on almost a third and Android accounting for two-thirds. [...]

Google's Chrome appears to be the most popular browser by far, accounting for 65.8 percent of all requests during 2024. Just 15.5 percent came from Apple's Safari browser, which leads the way on iOS devices, naturally. Microsoft's Edge accounted for 6.9 percent of browsing, while Mozilla Firefox stood at 4 percent. For search engines, Google also claimed the top spot, with a greater than 88 percent share of all search traffic that passed through Cloudflare. Yandex and Baidu were next with 3.1 percent and 2.7 percent, respectively, while Bing trailed with 2.6 percent. DuckDuckGo accounted for 0.9 percent of searches. You can read Cloudflare's full Year in Review here.