http://feed.informer.com/digests/IDAP6RXKUY/feeder Tech Digest 101 Respective post owners and feed distributors Fri, 21 Oct 2016 08:26:35 +0200 Feed Informer http://feed.informer.com/ https://arstechnica.com/ai/2025/04/metas-surprise-llama-4-drop-exposes-the-gap-between-ai-ambition-and-reality/ Ars Technica » Technology Lab urn:uuid:ac1c6976-656f-5fc5-2309-41918fc2b560 Mon, 07 Apr 2025 21:54:47 +0200 Touted 10M token context proves elusive, while early performance tests disappoint experts. <p>On Saturday, Meta <a href="https://ai.meta.com/blog/llama-4-multimodal-intelligence/">released</a> its newest Llama 4 multimodal AI models in a surprise weekend move that <a href="https://simonwillison.net/2025/Apr/5/llama-4-notes/">caught</a> some AI experts off guard. The announcement touted Llama 4 Scout and Llama 4 Maverick as major advancements, with Meta claiming top performance in their categories and an enormous 10 million token context window for Scout. But so far the open-weights models have received an initial mixed-to-negative reception from the AI community, highlighting a familiar tension between AI marketing and user experience.</p> <p>"The vibes around llama 4 so far are decidedly <a href="https://www.urbandictionary.com/define.php?term=mid">mid</a>," said independent AI researcher Simon Willison in a short interview with Ars Technica. Willison often <a href="https://simonwillison.net/2025/Apr/5/llama-4-notes/">checks the community pulse</a> around open source and open weights AI releases in particular.</p> <p>While Meta positions Llama 4 in competition with closed-model giants like OpenAI and Google, the company continues to use the term "open source" despite <a href="https://github.com/meta-llama/llama-models/blob/main/models/llama4/LICENSE">licensing restrictions</a> that prevent truly open use. As we have <a href="https://arstechnica.com/information-technology/2024/07/the-first-gpt-4-class-ai-model-anyone-can-download-has-arrived-llama-405b/">noted in the past</a> with previous Llama releases, "open weights" more accurately describes Meta's approach. Those who sign in and accept the license terms can download the two smaller Llama 4 models from <a href="https://huggingface.co/meta-llama">Hugging Face</a> or <a href="http://www.llama.com">llama.com</a>.</p><p><a href="https://arstechnica.com/ai/2025/04/metas-surprise-llama-4-drop-exposes-the-gap-between-ai-ambition-and-reality/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/04/metas-surprise-llama-4-drop-exposes-the-gap-between-ai-ambition-and-reality/#comments">Comments</a></p> https://www.youtube.com/watch?v=0ZedyrztolA Dave Lee urn:uuid:4c646bc4-3237-24b1-42ac-406dae7b83ff Sun, 06 Apr 2025 21:53:22 +0200 https://arstechnica.com/security/2025/04/nsa-warns-that-overlooked-botnet-technique-threatens-national-security/ Ars Technica » Technology Lab urn:uuid:a84713f5-0266-9c76-c91f-0c2c6fa8dc4e Fri, 04 Apr 2025 22:17:13 +0200 Used by nation-states and crime groups, fast flux bypasses many common defenses. <p>A technique that hostile nation-states and financially motivated ransomware groups are using to hide their operations poses a threat to critical infrastructure and national security, the National Security Agency has warned.</p> <p>The technique is known as fast flux. It allows decentralized networks operated by threat actors to hide their infrastructure and survive takedown attempts that would otherwise succeed. Fast flux works by cycling through a range of IP addresses and domain names that these botnets use to connect to the Internet. In some cases, IPs and domain names change every day or two; in other cases, they change almost hourly. The constant flux complicates the task of isolating the true origin of the infrastructure. It also provides redundancy. By the time defenders block one address or domain, new ones have already been assigned.</p> <h2>A significant threat</h2> <p>“This technique poses a significant threat to national security, enabling malicious cyber actors to consistently evade detection,” the NSA, FBI, and their counterparts from Canada, Australia, and New Zealand <a href="https://www.cisa.gov/news-events/cybersecurity-advisories/aa25-093a">warned Thursday</a>. “Malicious cyber actors, including cybercriminals and nation-state actors, use fast flux to obfuscate the locations of malicious servers by rapidly changing Domain Name System (DNS) records. Additionally, they can create resilient, highly available command and control (C2) infrastructure, concealing their subsequent malicious operations.”</p><p><a href="https://arstechnica.com/security/2025/04/nsa-warns-that-overlooked-botnet-technique-threatens-national-security/">Read full article</a></p> <p><a href="https://arstechnica.com/security/2025/04/nsa-warns-that-overlooked-botnet-technique-threatens-national-security/#comments">Comments</a></p> https://arstechnica.com/security/2025/04/are-new-google-e2ee-emails-really-end-to-end-encrypted-kinda-but-not-really/ Ars Technica » Technology Lab urn:uuid:dd756e79-a3c9-2fad-e19e-4c623b13b597 Thu, 03 Apr 2025 23:16:49 +0200 Yes, encryption/decryption occurs on end-user devices, but there's a catch. <p>When Google announced Tuesday that end-to-end encrypted messages were coming to Gmail for business users, some people balked, noting it wasn’t true E2EE as the term is known in privacy and security circles. Others wondered precisely how it works under the hood. Here’s a description of what the new service does and doesn’t do, as well as some of the basic security that underpins it.</p> <p>When Google uses the term E2EE in this context, it means that an email is encrypted inside Chrome, Firefox, or just about any other browser the sender chooses. As the message makes its way to its destination, it remains encrypted and can’t be decrypted until it arrives at its final destination, when it’s decrypted in the recipient's browser.</p> <h2>Giving S/MIME the heave-ho</h2> <p>The chief selling point of this new service is that it allows government agencies and the businesses that work with them to comply with a raft of security and privacy regulations and at the same time eliminates the massive headaches that have traditionally plagued anyone deploying such regulation-compliant email systems. Up to now, the most common means has been <a href="https://en.wikipedia.org/wiki/S/MIME">S/MIME</a>, a standard so complex and painful that only the bravest and most well-resourced organizations tend to implement it.</p><p><a href="https://arstechnica.com/security/2025/04/are-new-google-e2ee-emails-really-end-to-end-encrypted-kinda-but-not-really/">Read full article</a></p> <p><a href="https://arstechnica.com/security/2025/04/are-new-google-e2ee-emails-really-end-to-end-encrypted-kinda-but-not-really/#comments">Comments</a></p> https://arstechnica.com/information-technology/2025/04/ai-bots-strain-wikimedia-as-bandwidth-surges-50/ Ars Technica » Technology Lab urn:uuid:0133ebc6-67d8-1c80-edc0-ce904cdb5ca3 Wed, 02 Apr 2025 19:06:06 +0200 Automated AI bots seeking training data threaten Wikipedia project stability, foundation says. <p>On Tuesday, the Wikimedia Foundation <a href="https://diff.wikimedia.org/2025/04/01/how-crawlers-impact-the-operations-of-the-wikimedia-projects/">announced</a> that relentless AI scraping is putting strain on Wikipedia's servers. Automated bots seeking AI model training data for LLMs have been vacuuming up terabytes of data, growing the foundation's bandwidth used for downloading multimedia content by 50 percent since January 2024. It’s a scenario familiar across the free and open source software (FOSS) community, as we've <a href="https://arstechnica.com/ai/2025/03/devs-say-ai-crawlers-dominate-traffic-forcing-blocks-on-entire-countries/">previously detailed</a>.</p> <p>The Foundation hosts not only Wikipedia but also platforms like <a href="https://commons.wikimedia.org/wiki/Main_Page">Wikimedia Commons</a>, which offers 144 million media files under open licenses. For decades, this content has powered everything from search results to school projects. But since early 2024, AI companies have dramatically increased automated scraping through direct crawling, APIs, and bulk downloads to feed their hungry AI models. This exponential growth in non-human traffic has imposed steep technical and financial costs—often without the attribution that helps sustain Wikimedia’s volunteer ecosystem.</p> <p>The impact isn’t theoretical. The foundation says that when former US President Jimmy Carter died in December 2024, his Wikipedia page predictably drew millions of views. But the real stress came when users simultaneously streamed a 1.5-hour video of a 1980 debate from Wikimedia Commons. The surge doubled Wikimedia’s normal network traffic, temporarily maxing out several of its Internet connections. Wikimedia engineers quickly rerouted traffic to reduce congestion, but the event revealed a deeper problem: The baseline bandwidth had already been consumed largely by bots scraping media at scale.</p><p><a href="https://arstechnica.com/information-technology/2025/04/ai-bots-strain-wikimedia-as-bandwidth-surges-50/">Read full article</a></p> <p><a href="https://arstechnica.com/information-technology/2025/04/ai-bots-strain-wikimedia-as-bandwidth-surges-50/#comments">Comments</a></p> https://arstechnica.com/information-technology/2025/04/mcp-the-new-usb-c-for-ai-thats-bringing-fierce-rivals-together/ Ars Technica » Technology Lab urn:uuid:7946cad5-b8bb-6152-2d74-30478d2a2731 Tue, 01 Apr 2025 13:30:47 +0200 Model context protocol standardizes how AI uses data sources, supported by OpenAI and Anthropic. <p>What does it take to get OpenAI and Anthropic—two competitors in the AI assistant market—to get along? Despite a fundamental difference in direction that led Anthropic's founders to <a href="https://openai.com/index/organizational-update/">quit OpenAI</a> in 2020 and <a href="https://arstechnica.com/information-technology/2023/03/anthropic-introduces-claude-a-more-steerable-ai-competitor-to-chatgpt/">later create</a> the Claude AI assistant, a shared technical hurdle has now brought them together: How to easily connect their AI models to external data sources.</p> <p>The solution comes from Anthropic, which developed and released an open specification called <a href="https://www.anthropic.com/news/model-context-protocol">Model Context Protocol</a> (MCP) in November 2024. MCP establishes a royalty-free protocol that allows AI models to connect with outside data sources and services without requiring unique integrations for each service.</p> <p>"Think of MCP as a USB-C port for AI applications," <a href="https://modelcontextprotocol.io/introduction">wrote</a> Anthropic in MCP's documentation. The analogy is imperfect, but it represents the idea that, similar to how USB-C unified various cables and ports (with admittedly a <a href="https://learn.adafruit.com/understanding-usb-type-c-cable-types-pitfalls-and-more/cable-types-and-differences">debatable level</a> of success), MCP aims to standardize how AI models connect to the infoscape around them.</p><p><a href="https://arstechnica.com/information-technology/2025/04/mcp-the-new-usb-c-for-ai-thats-bringing-fierce-rivals-together/">Read full article</a></p> <p><a href="https://arstechnica.com/information-technology/2025/04/mcp-the-new-usb-c-for-ai-thats-bringing-fierce-rivals-together/#comments">Comments</a></p> https://arstechnica.com/tech-policy/2025/03/what-could-possibly-go-wrong-doge-to-rapidly-rebuild-social-security-codebase/ Ars Technica » Technology Lab urn:uuid:7c75408a-5d65-dd04-dee6-9b60f86ba571 Sat, 29 Mar 2025 15:08:49 +0100 A safe and proper rewrite should take years not months. <p>The so-called Department of Government Efficiency (DOGE) is starting to put together a team to migrate the Social Security Administration’s (SSA) computer systems entirely off one of its oldest programming languages in a matter of months, potentially putting the integrity of the system—and the benefits on which tens of millions of Americans rely—at risk.</p> <p>The project is being organized by Elon Musk lieutenant Steve Davis, multiple sources who were not given permission to talk to the media tell WIRED, and aims to migrate all SSA systems off COBOL, one of the first common business-oriented programming languages, and onto a more modern replacement like Java within a scheduled tight timeframe of a few months.</p> <p>Under any circumstances, a migration of this size and scale would be a massive undertaking, experts tell WIRED, but the expedited deadline runs the risk of obstructing payments to the more than <a href="https://www.ssa.gov/policy/docs/quickfacts/stat_snapshot/">65 million people</a> in the US currently receiving Social Security benefits.</p><p><a href="https://arstechnica.com/tech-policy/2025/03/what-could-possibly-go-wrong-doge-to-rapidly-rebuild-social-security-codebase/">Read full article</a></p> <p><a href="https://arstechnica.com/tech-policy/2025/03/what-could-possibly-go-wrong-doge-to-rapidly-rebuild-social-security-codebase/#comments">Comments</a></p> https://arstechnica.com/science/2025/03/scientists-are-storing-light-we-cannot-see-in-formats-meant-for-human-eyes/ Ars Technica » Technology Lab urn:uuid:9a06431d-b351-7762-fba8-82bdf1230675 Fri, 28 Mar 2025 23:11:00 +0100 New Spectral JPEG XL compression reduces file sizes, making spectral imaging more practical. <p>Imagine working with special cameras that capture light your eyes can't even see—ultraviolet rays that cause sunburn, infrared heat signatures that <a href="https://arstechnica.com/science/2024/09/new-multispectral-analysis-of-voynich-manuscript-reveals-hidden-details/">reveal hidden writing</a>, or specific wavelengths that plants use for photosynthesis. Or perhaps using a special camera designed to distinguish the subtle visible differences that make paint colors appear just right under specific lighting. Scientists and engineers do this every day, and they're drowning in the resulting data.</p> <p>A new compression format called <a href="https://jcgt.org/published/0014/01/04/">Spectral JPEG XL</a> might finally solve this growing problem in scientific visualization and computer graphics. Researchers Alban Fichet and Christoph Peters of Intel Corporation detailed the format in a <a href="https://jcgt.org/published/0014/01/04/">recent paper</a> published in the Journal of Computer Graphics Techniques (JCGT). It tackles a serious bottleneck for industries working with these specialized images. These spectral files can contain 30, 100, or more data points per pixel, causing file sizes to balloon into multi-gigabyte territory—making them unwieldy to store and analyze.</p> <p>When we think of digital images, we typically imagine files that store just three colors: red, green, and blue (RGB). This works well for everyday photos, but capturing the true color and behavior of light requires much more detail. <a href="https://en.wikipedia.org/wiki/Spectral_imaging">Spectral images</a> aim for this higher fidelity by recording light's intensity not just in broad RGB categories, but across dozens or even hundreds of narrow, specific wavelength bands. This detailed information primarily spans the visible spectrum and often extends into near-infrared and near-ultraviolet regions crucial for simulating how materials interact with light accurately.</p><p><a href="https://arstechnica.com/science/2025/03/scientists-are-storing-light-we-cannot-see-in-formats-meant-for-human-eyes/">Read full article</a></p> <p><a href="https://arstechnica.com/science/2025/03/scientists-are-storing-light-we-cannot-see-in-formats-meant-for-human-eyes/#comments">Comments</a></p> https://arstechnica.com/security/2025/03/oracle-is-mum-on-reports-it-has-experienced-2-separate-data-breaches/ Ars Technica » Technology Lab urn:uuid:4defc127-7df7-6e61-28e6-3002a9c6d2b5 Fri, 28 Mar 2025 20:41:14 +0100 Alleged breaches affect Oracle Cloud and Oracle Health. <p>Oracle isn’t commenting on recent reports that it has experienced two separate data breaches that have exposed sensitive personal information belonging to thousands of its customers.</p> <p>The most recent data breach report, <a href="https://www.bleepingcomputer.com/news/security/oracle-health-breach-compromises-patient-data-at-us-hospitals/">published Friday</a> by Bleeping Computer, said that Oracle Health—a health care software-as-a-service business the company acquired in 2022—had learned in February that a threat actor accessed one of its servers and made off with patient data from US hospitals. Bleeping Computer said Oracle Health customers have received breach notifications that were printed on plain paper rather than official Oracle letterhead and were signed by Seema Verma, the executive vice president &amp; GM of Oracle Health.</p> <p>The other report of a data breach occurred eight days ago, when an anonymous person using the handle rose87168 published a sampling of what they said were 6 million records of authentication data belonging to Oracle Cloud customers. Rose87168 <a href="https://www.bleepingcomputer.com/news/security/oracle-denies-data-breach-after-hacker-claims-theft-of-6-million-data-records/">told Bleeping Computer</a> that they had acquired the data a little more than a month earlier after exploiting a vulnerability that gave access to an Oracle Cloud server.</p><p><a href="https://arstechnica.com/security/2025/03/oracle-is-mum-on-reports-it-has-experienced-2-separate-data-breaches/">Read full article</a></p> <p><a href="https://arstechnica.com/security/2025/03/oracle-is-mum-on-reports-it-has-experienced-2-separate-data-breaches/#comments">Comments</a></p> https://arstechnica.com/security/2025/03/gemini-hackers-can-deliver-more-potent-attacks-with-a-helping-hand-from-gemini/ Ars Technica » Technology Lab urn:uuid:278d5cb7-8705-2952-e603-c3fc8d26df7e Fri, 28 Mar 2025 12:00:58 +0100 Hacking LLMs has always been more art than science. A new attack on Gemini could change that. <p>In the growing canon of AI security, the indirect prompt injection has emerged as the most powerful means for attackers to hack large language models such as OpenAI’s GPT-3 and GPT-4 or Microsoft’s Copilot. By exploiting a model's inability to distinguish between, on the one hand, developer-defined prompts and, on the other, text in external content LLMs interact with, indirect prompt injections are remarkably effective at invoking harmful or otherwise unintended actions. Examples include divulging end users’ confidential contacts or emails and delivering falsified answers that have the potential to corrupt the integrity of important calculations.</p> <p>Despite the power of prompt injections, attackers face a fundamental challenge in using them: The inner workings of so-called closed-weights models such as GPT, Anthropic’s Claude, and Google’s Gemini are closely held secrets. Developers of such proprietary platforms tightly restrict access to the underlying code and training data that make them work and, in the process, make them black boxes to external users. As a result, devising working prompt injections requires labor- and time-intensive trial and error through redundant manual effort.</p> <h2>Algorithmically generated hacks</h2> <p>For the first time, academic researchers have devised a means to create computer-generated prompt injections against Gemini that have much higher success rates than manually crafted ones. The new method abuses fine-tuning, a feature offered by some closed-weights models for training them to work on large amounts of private or specialized data, such as a law firm’s legal case files, patient files or research managed by a medical facility, or architectural blueprints. Google makes its <a href="https://ai.google.dev/gemini-api/docs/model-tuning">fine-tuning for Gemini’s API</a> available free of charge.</p><p><a href="https://arstechnica.com/security/2025/03/gemini-hackers-can-deliver-more-potent-attacks-with-a-helping-hand-from-gemini/">Read full article</a></p> <p><a href="https://arstechnica.com/security/2025/03/gemini-hackers-can-deliver-more-potent-attacks-with-a-helping-hand-from-gemini/#comments">Comments</a></p> https://www.youtube.com/watch?v=zBk8kAqPzMo Dave Lee urn:uuid:a5939407-b36b-80fb-cb0a-52b50dccb2dc Thu, 27 Mar 2025 21:30:52 +0100 https://arstechnica.com/ai/2025/03/openais-new-ai-image-generator-is-potent-and-bound-to-provoke/ Ars Technica » Technology Lab urn:uuid:badf11f4-bd31-4448-6e76-5e992ef99406 Thu, 27 Mar 2025 12:15:10 +0100 The visual apocalypse is probably nigh, but perhaps seeing was never believing. <p>The arrival of OpenAI's DALL-E 2 in the spring of 2022 marked a turning point in AI when text-to-image generation suddenly became accessible to a select group of users, creating a <a href="https://arstechnica.com/information-technology/2024/04/when-ai-images-were-mind-blowing-early-users-recall-the-first-days-of-dall-e-2/">community of digital explorers</a> who experienced wonder and controversy as the technology automated the act of visual creation.</p> <p>But like many early AI systems, DALL-E 2 struggled with consistent text rendering, often producing garbled words and phrases within images. It also had limitations in following complex prompts with multiple elements, sometimes missing key details or misinterpreting instructions. These shortcomings left room for improvement that OpenAI would address in subsequent iterations, such as <a href="https://arstechnica.com/information-technology/2023/11/from-toy-to-tool-dall-e-3-is-a-wake-up-call-for-visual-artists-and-the-rest-of-us/">DALL-E 3</a> in 2023.</p> <p>On Tuesday, OpenAI <a href="https://openai.com/index/introducing-4o-image-generation/">announced</a> new multimodal image generation capabilities that are directly integrated into its GPT-4o AI language model, making it the default image generator within the ChatGPT interface. The integration, called "4o Image Generation" (which we'll call "4o IG" for short), allows the model to follow prompts more accurately (with better text rendering than DALL-E 3) and respond to chat context for image modification instructions.</p><p><a href="https://arstechnica.com/ai/2025/03/openais-new-ai-image-generator-is-potent-and-bound-to-provoke/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/openais-new-ai-image-generator-is-potent-and-bound-to-provoke/#comments">Comments</a></p> https://arstechnica.com/information-technology/2025/03/broadcoms-vmware-says-siemens-pirated-thousands-of-copies-of-its-software/ Ars Technica » Technology Lab urn:uuid:78b20595-b6df-1b27-7511-eb762a8731dc Wed, 26 Mar 2025 15:54:33 +0100 VMware claims Siemens showed it a list of the VMware products it's using unlicensed. <p>VMware is suing the US arm of industrial giant AG Siemens. The Broadcom company claims that Siemens outed itself by revealing to VMware that it downloaded and distributed multiple copies of VMware products without buying a license.</p> <p>VMware filed the lawsuit (<a href="https://cdn.arstechnica.net/wp-content/uploads/2025/03/Vmware_vs_siemens.pdf">PDF</a>) on March 21 in the US District Court for the District of Delaware, as spotted by <a href="https://www.theregister.com/2025/03/26/vmware_sues_siemens_for_using/">The Register</a>.</p> <p>In the complaint, VMware says that it has had a Master Software License and Service Agreement with Siemens since November 28, 2012. The virtualization company claims that in September, Siemens sent VMware a purchase order for maintenance and support services. Siemens was reportedly looking to exercise a previously agreed-upon option for a one-year renewal of support services. However, the list of VMware technology that Siemens was seeking support for "included a large number of products for which [VMware] had no record of Siemens AG purchasing a license," the complaint says.</p><p><a href="https://arstechnica.com/information-technology/2025/03/broadcoms-vmware-says-siemens-pirated-thousands-of-copies-of-its-software/">Read full article</a></p> <p><a href="https://arstechnica.com/information-technology/2025/03/broadcoms-vmware-says-siemens-pirated-thousands-of-copies-of-its-software/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/devs-say-ai-crawlers-dominate-traffic-forcing-blocks-on-entire-countries/ Ars Technica » Technology Lab urn:uuid:c77fa20d-07f9-dcd8-12ab-592022a7d2b7 Tue, 25 Mar 2025 22:36:58 +0100 AI bots hungry for data are taking down FOSS sites by accident, but humans are fighting back. <p>Software developer Xe Iaso <a href="https://xeiaso.net/notes/2025/amazon-crawler/">reached a breaking point</a> earlier this year when aggressive AI crawler traffic from Amazon overwhelmed their Git repository service, repeatedly causing instability and downtime. Despite configuring standard defensive measures—adjusting robots.txt, blocking known crawler user-agents, and filtering suspicious traffic—Iaso found that AI crawlers continued evading all attempts to stop them, spoofing user-agents and cycling through residential IP addresses as proxies.</p> <p>Desperate for a solution, Iaso eventually resorted to moving their server behind a VPN and creating "Anubis," a custom-built proof-of-work challenge system that forces web browsers to solve computational puzzles before accessing the site. "It's futile to block AI crawler bots because they lie, change their user agent, use residential IP addresses as proxies, and more," Iaso wrote in a <a href="https://xeiaso.net/notes/2025/amazon-crawler/">blog post</a> titled "a desperate cry for help." "I don't want to have to close off my Gitea server to the public, but I will if I have to."</p> <p>Iaso's story highlights a broader crisis rapidly spreading across the open source community, as what appear to be aggressive AI crawlers increasingly overload community-maintained infrastructure, causing what amounts to persistent distributed denial-of-service (DDoS) attacks on vital public resources. According to a <a href="https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/">comprehensive recent report</a> from LibreNews, some open source projects now see as much as 97 percent of their traffic originating from AI companies' bots, dramatically increasing bandwidth costs, service instability, and burdening already stretched-thin maintainers.</p><p><a href="https://arstechnica.com/ai/2025/03/devs-say-ai-crawlers-dominate-traffic-forcing-blocks-on-entire-countries/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/devs-say-ai-crawlers-dominate-traffic-forcing-blocks-on-entire-countries/#comments">Comments</a></p> https://www.youtube.com/watch?v=iKcu4o9rjfA Dave Lee urn:uuid:b775e88f-4856-cc84-c9f3-026feaee68b6 Tue, 25 Mar 2025 14:54:55 +0100 https://arstechnica.com/information-technology/2025/03/europe-is-looking-for-alternatives-to-us-cloud-providers/ Ars Technica » Technology Lab urn:uuid:f83adfad-0b01-604c-53c6-a11810a20d5a Tue, 25 Mar 2025 14:12:52 +0100 Some European cloud companies have seen an increase in business. <p>The global backlash against the second Donald Trump <a href="https://www.wired.com/story/federal-auditors-doge-elon-musk/">administration</a> keeps on growing. Canadians have <a href="https://thelogic.co/intelligence/subscriber-survey/buy-canadian/">boycotted US-made products</a>, anti–Elon Musk posters have <a href="https://www.independent.co.uk/news/uk/home-news/elon-musk-protests-tesla-takedown-london-b2716372.html">appeared across London</a> amid widespread <a href="https://www.wired.com/story/elon-musks-toxicity-could-spell-disaster-for-tesla/">Tesla protests</a>, and European officials have <a href="https://ec.europa.eu/commission/presscorner/detail/en/ip_25_793">drastically increased</a> military spending as US support for Ukraine falters. Dominant US tech services may be the next focus.</p> <p>There are early signs that some European companies and governments are souring on their use of American cloud services provided by the three so-called hyperscalers. Between them, Google Cloud, Microsoft Azure, and Amazon Web Services (AWS) host vast swathes of the Internet and keep thousands of businesses running. However, some organizations appear to be reconsidering their use of these companies’ cloud services—including servers, storage, and databases—citing uncertainties around privacy and data access fears under the Trump administration.</p> <p>“There’s a huge appetite in Europe to de-risk or decouple the over-dependence on US tech companies, because there is a concern that they could be weaponized against European interests,” says Marietje Schaake, a nonresident fellow at Stanford’s Cyber Policy Center and a former decadelong member of the European Parliament.</p><p><a href="https://arstechnica.com/information-technology/2025/03/europe-is-looking-for-alternatives-to-us-cloud-providers/">Read full article</a></p> <p><a href="https://arstechnica.com/information-technology/2025/03/europe-is-looking-for-alternatives-to-us-cloud-providers/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/you-can-now-download-the-source-code-that-sparked-the-ai-boom/ Ars Technica » Technology Lab urn:uuid:511d318e-789f-69d9-81df-5224aaaf6c4b Mon, 24 Mar 2025 23:14:15 +0100 CHM releases code for 2012 AlexNet breakthrough that proved "deep learning" could work. <p>On Thursday, Google and the Computer History Museum (CHM) <a href="https://computerhistory.org/blog/chm-releases-alexnet-source-code/">jointly released</a> the source code for <a href="https://en.wikipedia.org/wiki/AlexNet">AlexNet</a>, the convolutional neural network (CNN) that many credit with transforming the AI field in 2012 by proving that "deep learning" could achieve things conventional AI techniques could not.</p> <p><a href="https://arstechnica.com/science/2019/12/how-neural-networks-work-and-why-theyve-become-a-big-business/">Deep learning</a>, which uses multi-layered neural networks that can learn from data without explicit programming, represented a significant departure from traditional AI approaches that relied on hand-crafted rules and features.</p> <p>The Python code, now available on CHM's <a href="https://github.com/computerhistory/AlexNet-Source-Code">GitHub page</a> as open source software, offers AI enthusiasts and researchers a glimpse into a key moment of computing history. AlexNet served as a watershed moment in AI because it could accurately identify objects in photographs with unprecedented accuracy—correctly classifying images into one of 1,000 categories like "strawberry," "school bus," or "golden retriever" with significantly fewer errors than previous systems.</p><p><a href="https://arstechnica.com/ai/2025/03/you-can-now-download-the-source-code-that-sparked-the-ai-boom/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/you-can-now-download-the-source-code-that-sparked-the-ai-boom/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/cloudflare-turns-ai-against-itself-with-endless-maze-of-irrelevant-facts/ Ars Technica » Technology Lab urn:uuid:9c57cc53-e9a1-1a9b-2626-7fa9fcbb19bc Fri, 21 Mar 2025 22:14:35 +0100 New approach punishes AI companies that ignore "no crawl" directives. <p>On Wednesday, web infrastructure provider Cloudflare announced a new feature called "<a href="https://blog.cloudflare.com/ai-labyrinth/">AI Labyrinth</a>" that aims to combat unauthorized AI data scraping by serving fake AI-generated content to bots. The tool will attempt to thwart AI companies that crawl websites without permission to collect training data for large language models that power AI assistants like <a href="https://arstechnica.com/information-technology/2023/11/chatgpt-was-the-spark-that-lit-the-fire-under-generative-ai-one-year-ago-today/">ChatGPT</a>.</p> <p>Cloudflare, founded in 2009, is probably best known as a company that <a href="https://en.wikipedia.org/wiki/Cloudflare">provides</a> infrastructure and security services for websites, particularly protection against <a href="https://en.wikipedia.org/wiki/DDoS_mitigation">distributed denial-of-service</a> (DDoS) attacks and other malicious traffic.</p> <p class="whitespace-pre-wrap break-words">Instead of simply blocking bots, Cloudflare's new system lures them into a "maze" of realistic-looking but irrelevant pages, wasting the crawler's computing resources. The approach is a notable shift from the standard block-and-defend strategy used by most website protection services. Cloudflare says blocking bots sometimes backfires because it alerts the crawler's operators that they've been detected.</p><p><a href="https://arstechnica.com/ai/2025/03/cloudflare-turns-ai-against-itself-with-endless-maze-of-irrelevant-facts/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/cloudflare-turns-ai-against-itself-with-endless-maze-of-irrelevant-facts/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/anthropics-new-ai-search-feature-digs-through-the-web-for-answers/ Ars Technica » Technology Lab urn:uuid:4111bd18-7dc1-2c9d-145b-dad8137c1f5c Fri, 21 Mar 2025 20:08:56 +0100 Anthropic Claude just caught up with a ChatGPT feature from 2023—but will it be accurate? <p>On Thursday, Anthropic <a href="https://www.anthropic.com/news/web-search">introduced</a> web search capabilities for its AI assistant Claude, enabling the assistant to access current information online. Previously, the latest AI model that powers Claude could only rely on data absorbed during its neural network training process, having a "knowledge cutoff" of October 2024.</p> <p>Claude's web search is currently available in feature preview for paid users in the United States, with plans to expand to free users and additional countries in the future. After users enable the feature in their profile settings, Claude will automatically determine when to use web search to answer a query or find more recent information.</p> <p>The new feature works with <a href="https://arstechnica.com/ai/2025/02/claude-3-7-sonnet-debuts-with-extended-thinking-to-tackle-complex-problems/">Claude 3.7 Sonnet</a> and requires a paid subscription. The addition brings Claude in line with competitors like Microsoft Copilot and ChatGPT, which already offer similar functionality. ChatGPT <a href="https://arstechnica.com/information-technology/2023/03/chatgpt-gets-eyes-and-ears-with-plugins-that-can-interface-ai-with-the-world/">first added</a> the ability to grab web search results as a plugin in March 2023, so this new feature is a long time coming.</p><p><a href="https://arstechnica.com/ai/2025/03/anthropics-new-ai-search-feature-digs-through-the-web-for-answers/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/anthropics-new-ai-search-feature-digs-through-the-web-for-answers/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/ai-beats-humans-at-meme-humor-but-the-best-joke-is-still-human-made/ Ars Technica » Technology Lab urn:uuid:1b82763d-7779-7da1-cafa-3a54484ab528 Wed, 19 Mar 2025 23:12:47 +0100 Mollick proclaims "the meme Turing Test has been passed," but a new study offers a key caveat. <p>A <a href="https://arxiv.org/abs/2501.11433">new study</a> examining meme creation found that AI-generated meme captions on existing famous meme images scored higher on average for humor, creativity, and "shareability" than those made by people. Even so, people still created the most exceptional individual examples.</p> <p>The research, which will be presented at the <a href="https://iui.acm.org/2025/">2025 International Conference on Intelligent User Interfaces</a>, reveals a nuanced picture of how AI and humans perform differently in humor creation tasks. Still, the results were surprising enough to have one expert declaring victory for the machines.</p> <p>"I regret to announce that the meme Turing Test has been passed," <a href="https://bsky.app/profile/emollick.bsky.social/post/3lkjvtg6wmk2i">wrote</a> Wharton professor Ethan Mollick on Bluesky after reviewing the study results. Mollick studies AI academically, and he's referring to a <a href="https://en.wikipedia.org/wiki/Turing_test">famous test</a> proposed by computing pioneer Alan Turing in 1950 that seeks to determine whether humans can distinguish between AI outputs and human-created content.</p><p><a href="https://arstechnica.com/ai/2025/03/ai-beats-humans-at-meme-humor-but-the-best-joke-is-still-human-made/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/ai-beats-humans-at-meme-humor-but-the-best-joke-is-still-human-made/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/nvidia-announces-dgx-desktop-personal-ai-supercomputers/ Ars Technica » Technology Lab urn:uuid:b3ca0dec-2946-36b0-b504-c0bf9fcafe6d Tue, 18 Mar 2025 22:19:53 +0100 Asus, Dell, HP and others to produce powerful desktop machines that run AI models locally. <p>During Tuesday's Nvidia GTX keynote, CEO Jensen Huang unveiled two "personal AI supercomputers" called DGX Spark and DGX Station, both powered by the Grace Blackwell platform. In a way, they are a new type of AI PC architecture specifically built for running neural networks, and five major PC manufacturers will build the supercomputers.</p> <p>These desktop systems, first previewed as "<a href="https://arstechnica.com/ai/2025/01/nvidias-first-desktop-pc-can-run-local-ai-models-for-3000/">Project DIGITS</a>" in January, aim to bring AI capabilities to developers, researchers, and data scientists who need to prototype, fine-tune, and run large AI models locally. DGX systems can serve as standalone desktop AI labs or "bridge systems" that allow AI developers to move their models from desktops to DGX Cloud or any AI cloud infrastructure with few code changes.</p> <p>Huang explained the rationale behind these new products in a <a href="https://nvidianews.nvidia.com/news/nvidia-announces-dgx-spark-and-dgx-station-personal-ai-computers">news release</a>, saying, "AI has transformed every layer of the computing stack. It stands to reason a new class of computers would emerge—designed for AI-native developers and to run AI-native applications."</p><p><a href="https://arstechnica.com/ai/2025/03/nvidia-announces-dgx-desktop-personal-ai-supercomputers/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/nvidia-announces-dgx-desktop-personal-ai-supercomputers/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/nvidia-announces-rubin-ultra-and-feynman-ai-chips-for-2027-and-2028/ Ars Technica » Technology Lab urn:uuid:3ce7c0b4-50a5-441d-664d-a3556aae2b1e Tue, 18 Mar 2025 22:10:57 +0100 CEO Jensen Huang says new chips will power robots and billions of AI agents. <p>On Tuesday at Nvidia's GTC 2025 conference in San Jose, California, CEO Jensen Huang <a href="https://www.nvidia.com/gtc/keynote/">revealed</a> several new AI-accelerating GPUs the company plans to release over the coming months and years. He also revealed more specifications about previously announced chips.</p> <p>The centerpiece announcement was Vera Rubin, <a href="https://arstechnica.com/information-technology/2024/06/nvidia-jumps-ahead-of-itself-and-reveals-next-gen-rubin-ai-chips-in-keynote-tease/">first teased</a> at Computex 2024 and now scheduled for release in the second half of 2026. This GPU, named after a <a href="https://en.wikipedia.org/wiki/Vera_Rubin">famous astronomer</a>, will feature tens of terabytes of memory and comes with a custom Nvidia-designed CPU called Vera.</p> <p>According to Nvidia, Vera Rubin will deliver significant performance improvements over its predecessor, <a href="https://arstechnica.com/information-technology/2024/03/nvidia-unveils-blackwell-b200-the-worlds-most-powerful-chip-designed-for-ai/">Grace Blackwell</a>, particularly for AI training and inference.</p><p><a href="https://arstechnica.com/ai/2025/03/nvidia-announces-rubin-ultra-and-feynman-ai-chips-for-2027-and-2028/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/nvidia-announces-rubin-ultra-and-feynman-ai-chips-for-2027-and-2028/#comments">Comments</a></p> https://arstechnica.com/ai/2025/03/farewell-photoshop-googles-new-ai-lets-you-edit-images-by-asking/ Ars Technica » Technology Lab urn:uuid:2b2223bc-5205-52b0-da72-21f2f24c150e Tue, 18 Mar 2025 12:15:35 +0100 New AI allows no-skill photo editing, including adding objects and removing watermarks. <p>There's a new Google AI model in town, and it can generate or edit images as easily as it can create text—as part of its chatbot conversation. The results aren't perfect, but it's quite possible everyone in the near future will be able to manipulate images this way.</p> <p>Last Wednesday, Google <a href="https://developers.googleblog.com/en/experiment-with-gemini-20-flash-native-image-generation/">expanded access</a> to Gemini 2.0 Flash's native image-generation capabilities, making the experimental feature available to anyone using <a href="https://aistudio.google.com/prompts/new_chat">Google AI Studio</a>. Previously limited to testers since December, the multimodal technology integrates both native text and image processing capabilities into one AI model.</p> <p>The new model, titled "Gemini 2.0 Flash (Image Generation) Experimental," flew somewhat under the radar last week, but it has been garnering more attention over the past few days due to its ability to <a href="https://techcrunch.com/2025/03/16/people-are-using-googles-new-ai-model-to-remove-watermarks-from-images/">remove watermarks</a> from images, albeit with artifacts and a reduction in image quality.</p><p><a href="https://arstechnica.com/ai/2025/03/farewell-photoshop-googles-new-ai-lets-you-edit-images-by-asking/">Read full article</a></p> <p><a href="https://arstechnica.com/ai/2025/03/farewell-photoshop-googles-new-ai-lets-you-edit-images-by-asking/#comments">Comments</a></p> https://www.youtube.com/watch?v=J4qwuCXyAcU Dave Lee urn:uuid:29592cc8-bb18-da2c-30b5-a0aebae22990 Tue, 11 Mar 2025 14:11:17 +0100 https://www.youtube.com/watch?v=w-WhZT02ubg Dave Lee urn:uuid:760b6893-91bf-fef0-e2b7-f1666b8e3f7a Thu, 27 Feb 2025 12:30:35 +0100 https://www.youtube.com/watch?v=IVbm2a6lVBo Dave Lee urn:uuid:e9ba08aa-9c47-12f3-bbcb-4be99efa96d5 Tue, 18 Feb 2025 15:09:43 +0100 https://www.youtube.com/watch?v=931n_5Usfiw Dave Lee urn:uuid:754f4743-ea17-8896-92e7-a99d75c5aed8 Tue, 11 Feb 2025 16:40:05 +0100 https://www.youtube.com/watch?v=kE6nw47Q80E Dave Lee urn:uuid:9cc935ae-f714-1f7d-d8eb-4d4f3dfc7acd Wed, 22 Jan 2025 19:19:31 +0100 https://www.youtube.com/watch?v=8njaN9ZaSdA Dave Lee urn:uuid:71098191-a188-6815-3861-272c38df3958 Tue, 21 Jan 2025 20:09:42 +0100 https://www.youtube.com/watch?v=NFvOMwfRkaM Dave Lee urn:uuid:f57f5e74-9d9a-0d5b-f351-07bf7b661c0e Sat, 18 Jan 2025 20:25:15 +0100 https://www.youtube.com/watch?v=-xk7-43lX-c Dave Lee urn:uuid:eab783bf-156b-bc86-b863-c3ccdc7ef07d Thu, 09 Jan 2025 18:09:49 +0100 https://www.youtube.com/watch?v=yMHNrCiQboI Dave Lee urn:uuid:0f971db4-bfe5-ff83-55a4-e41d0343c340 Tue, 07 Jan 2025 19:08:31 +0100 https://www.youtube.com/watch?v=BFAkZJsZn6k Dave Lee urn:uuid:7683b2d5-f79f-ceb6-fbb7-0f07bcb7f667 Tue, 07 Jan 2025 17:34:56 +0100 https://www.youtube.com/watch?v=Iq1JeXKYg5k Dave Lee urn:uuid:cbe032b4-285d-8fb7-18ae-b2b1bbe8d941 Tue, 07 Jan 2025 08:05:49 +0100 https://www.youtube.com/watch?v=nZlhI-L_Egk Dave Lee urn:uuid:352d8376-3b7c-53d4-cef0-5aacee810f4a Thu, 26 Dec 2024 07:55:36 +0100 https://www.cnet.com/reviews/sling-tv-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:b210b42b-efed-a45c-a2c4-d087f6b3aa28 Thu, 01 Aug 2024 06:00:00 +0200 Sling TV Blue offers cord-cutters a wealth of live channels for an affordable price, but you may need to bring your own antenna. https://www.cnet.com/roadshow/reviews/2022-gmc-hummer-ev-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:9fd90772-44cb-ca1d-33da-e6c11bf98225 Mon, 02 Jan 2023 11:00:01 +0100 After the allure of the Hummer's physics-defying acceleration wears off, there isn't a whole lot left to love. https://www.cnet.com/reviews/amazon-fire-tv-stick-lite-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:73dfc366-45d8-9c18-d47a-35edc9fb5a5f Mon, 19 Dec 2022 19:01:00 +0100 The first ultrabudget streamer with a voice remote talks circles around Roku. https://www.cnet.com/roadshow/reviews/2023-land-rover-range-rover-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:5dd95bad-862a-32b2-450c-0a3ae3722e28 Wed, 14 Dec 2022 11:00:01 +0100 The latest generation of the ubiquitous luxury SUV is damn near perfect. https://www.cnet.com/reviews/roku-streambar-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:80392a32-0fa1-c196-1a2c-8a86bac4e712 Sun, 04 Dec 2022 12:00:00 +0100 Editor's Choice: This tiny bar adds great 4K HDR streaming, and solid sound, to any TV. https://www.cnet.com/reviews/anker-nebula-mars-ii-pro-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:cea1a7d9-6072-f0e7-bf18-f568466178b0 Thu, 01 Dec 2022 16:31:00 +0100 Editor's Choice: This tiny, battery-powered entertainment machine does a lot of things right. https://www.cnet.com/reviews/benq-cinehome-ht2050a-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:b5e2e90e-d281-8704-679d-59d71053f63b Thu, 01 Dec 2022 01:00:00 +0100 Editor's Choice: Excellent contrast, accurate color and solid features make it our favorite sub-$1,000 projector for 2020. https://www.cnet.com/reviews/epson-home-cinema-5050ub-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:c7ae4ced-4857-5f52-07bf-f3c65a30e445 Wed, 30 Nov 2022 19:21:00 +0100 Editor's Choice: A higher-end projector worthy of the "home cinema" moniker. https://www.cnet.com/roadshow/reviews/2023-bmw-ix-m60-sports-activity-vehicle-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:b9bbe11b-2dca-90dd-f0fe-7bc952eaca49 Thu, 24 Nov 2022 12:00:01 +0100 The more powerful spec of BMW's iX electric SUV is certainly quicker than the base model, but that extra oomph comes at too high a cost. https://www.cnet.com/reviews/lg-oled-c2-series-2022-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:7b10c6da-620c-fb57-f710-0d0ff5399fd5 Tue, 15 Nov 2022 16:53:00 +0100 The C2 sets the pace for 2022 with superb picture quality, numerous sizes and a track record of success. https://www.cnet.com/reviews/blink-mini-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:5e1c698c-24ea-f2cb-40ac-3409866cb4a6 Fri, 21 Oct 2022 22:13:00 +0200 Available in black or white, Amazon's Blink Mini sports many features for a reasonable price. https://www.cnet.com/roadshow/reviews/2023-bmw-i4-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:b1b3b865-078d-5427-ba62-047faca3b280 Fri, 14 Oct 2022 21:33:00 +0200 You don't need to pay more for dual-motor i4 M50 to have a good time, but it's hard to argue with more power and fun. https://www.cnet.com/roadshow/reviews/2022-toyota-tundra-4wd-trd-pro-hybrid-crewmax-5-5-bed-3-5l-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:72c40ca9-222d-855b-9cd6-1aca83a0ec75 Tue, 11 Oct 2022 11:00:01 +0200 Toyota's off-road-ready pickup is mighty fine in daily driving, too. https://www.cnet.com/roadshow/reviews/2023-cadillac-xt6-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:ff9bc813-a50e-a023-686f-29f292247676 Thu, 22 Sep 2022 11:00:00 +0200 The addition of Caddy's hands-free highway driving assistant makes the XT6 one of the most interesting luxury SUVs on the road -- but there's a catch. https://www.cnet.com/roadshow/reviews/2022-chevrolet-silverado-1500-4wd-crew-cab-157-lt-trail-boss-review/#ftag=CADe9e329a CNET Reviews - Most Recent Reviews urn:uuid:a272f538-1e44-3318-677e-dd4072b500b7 Thu, 15 Sep 2022 11:00:01 +0200 This rough-and-ready pickup is home to Chevy's new Android-based, Google Assistant-powered dashboard tech.