Nvidia plans to produce a simplified, more affordable version of its Blackwell GPU for the Chinese market starting in June. Based on the RTX Pro 6000D and priced between $6,500 and $8,000, this new …

Google has expanded its AI video model Veo 3 to over 70 countries shortly after launch, excluding the EU for now. Josh Woodward, VP of Gemini at Google, announced the rollout on platform X. Veo 3 …

Publishers criticize Google’s new AI Mode feature, which allows chatbot-like interactions in search, as theft. The News/Media Alliance (NMA), representing nearly 2,000 media members in the US …

Anthropic has activated its highest safety level—AI Safety Level 3—for the new Claude 4 Opus model. This real-time monitoring aims to block potentially dangerous outputs, officially due to the model’s …

Anthropic has launched Claude 4, introducing the Sonnet 4 and Opus 4 models after a year-long pause. Aimed at the emerging “agentic era,” Opus 4 can handle multi-hour tasks autonomously—successfully …

OpenAI plans to move from nonprofit control to a profit-oriented Public Benefit Corporation structure, aiming to balance public interest with investor returns. While the company cites the need for …

OpenAI now offers its latest image generation model, gpt-image-1, through an official API, allowing developers to embed customizable image creation into their own apps and platforms. The model, …

xAI has added visual capabilities to its chatbot Grok through a new feature called Grok Vision. Available first on iPhones, the update allows users to point their phone at objects, signs, or documents …

OpenAI has revamped its safety framework to better manage the risks of advanced AI systems. The updated approach focuses on identifying and mitigating capabilities that could lead to serious, …

OpenAI has introduced two new models, o3 and o4-mini, as part of its o-series, showcasing significant advancements in autonomous tool use and visual reasoning. These models can independently decide …

Microsoft has rolled out its visual AI assistant, Copilot Vision, to the Edge browser. The feature allows users to interact with web content through natural language, though it’s currently limited to …

According to The Verge, OpenAI is developing an AI-powered social media platform that could rival Elon Musk’s X. Spearheaded by ex-TikTok and Snap product designer Johnny Lu, the project appears to be …

Alibaba’s Quark app has surged to the top of China’s AI market, reaching around 150 million monthly active users and surpassing rivals like ByteDance’s Doubao and DeepSeek. Originally a cloud storage …

Slopsquatting is an emerging security threat that exploits hallucinated package names in AI-generated code. Coined by researcher Seth Larson, the attack involves malicious actors uploading harmful …

OpenAI has unveiled GPT-4.1, the latest version of its flagship model, combining the strengths of GPT-4 with significant upgrades. Available to ChatGPT Plus users, GPT-4.1 offers improved reasoning, …

Apple is reportedly lagging behind in AI development due to an internal decision that limited critical GPU acquisitions. According to the New York Times, Apple CFO Luca Maestri blocked a large GPU …

Elon Musk’s AI company xAI has modified its chatbot Grok to avoid making critical statements about Musk. Earlier versions had listed him as a major spreader of misinformation, but Grok 2 and 3 now …

OpenAI is accelerating the release of new AI models, significantly reducing safety testing time, according to the Financial Times. While GPT-4 underwent six months of risk evaluation, its successor …

Google has unveiled DolphinGemma, an AI model designed to interpret and generate dolphin sounds, aiming to enable basic communication with dolphins. Developed with Georgia Tech and the Wild Dolphin …

Meta hofft auf einen „Knock-on-Effekt“ durch die Integration von Meta AI in Produkte wie WhatsApp und Instagram, um die KI-Nutzung in Europa zu fördern. Laut Public Policy Director Semjon Rens ist …

The focus of criticism isn’t necessarily the laws themselves, like the AI Act or GDPR, but rather the slow and complex regulatory processes involved. While the AI Act addresses general-purpose …

Introducing AI products in Europe requires navigating a highly complex legal landscape, with over 270 EU regulators plus 18 national data protection authorities involved. Companies like Meta face a …

Due to complex EU regulations, Meta AI is only partially available in Germany and not yet fully launched across the EU. For instance, features like integration with Ray-Ban smart glasses are still …

A report by the NGO Statewatch has raised concerns over the UK Justice Ministry’s use of algorithms to predict potential murders. The system analyzes sensitive data from up to 500,000 individuals, …

The State of AI Index 2025 by Stanford’s Institute for Human-Centered AI highlights AI’s growing role in society and business. For instance, 223 AI-powered medical devices were approved by the …

Anthropic has introduced a new premium subscription for its Claude model called the “Max Plan.” Targeted at heavy users, it offers significantly higher usage limits and priority access to …

WordPress has introduced a free AI-powered website builder aimed at small businesses, bloggers, and creators—though hosting may incur additional costs. While not suited for e-commerce, the tool …

The EU aims to close the AI infrastructure gap with its “European AI Continent” strategy, proposing up to €200 billion in funding—€20 billion of which would go toward building up to five AI …

At its Cloud Next conference, Google introduced major AI advancements, including the powerful new Ironwood chip—potentially surpassing the world’s fastest supercomputers—and Gemini 2.5 Flash, a …

Nvidia’s research team has developed a new method for creating AI-generated videos with coherent stories up to one minute long. Previous models like Sora or Veo 2 were limited to 20 seconds due to the …

GPT-4o’s new image generator stands out for its exceptional image quality, precise text integration, and detailed template replication. Unlike other models like Midjourney or DALL-E, it may use an …

Google DeepMind is offering employees in the UK paid sabbaticals rather than letting them join competitors like OpenAI or Meta. According to Business Insider, employee contracts include strict …

Google’s AI research tool, NotebookLM, currently available only through the browser, will soon launch as a mobile app for Android and iOS. While the web interface suits it well, a native desktop …

Google’s new whitepaper suggests generative AI could reduce global emissions by 5–10% by 2030—equal to the EU’s annual footprint. The plan focuses on three areas: enabling data and skills, deploying …

Amazon’s new Nova Sonic voice model rivals OpenAI and Google in speech quality and speed, while costing around 80% less than GPT-4o’s voice mode. It’s available via Amazon Bedrock and already …

In its Economic Blueprint, OpenAI calls for a major reset of the EU’s AI strategy, centered on four principles: better infrastructure, reduced bureaucracy, deeper AI integration in daily life, and …

OpenAI is reportedly interested in acquiring io Products, a stealth startup from former Apple design chief Jony Ive’s company, LoveFrom. The startup is developing a screenless AI device, with OpenAI …

Microsoft has integrated the Llama 4 Scout and Maverick models into Azure AI Foundry and Azure Databricks. These models come with customizable safety features, like data filtering, to protect …

Meta allegedly inflated the benchmark results for its Llama 4 models. In a blog post, they claimed an ELO score of 1417, surpassing GPT-4 and just below Google’s Gemini 2.5 Pro. However, testers …

Google’s Gemini Live lets Android users share their camera and screen with the AI, enabling smarter help with things like tech support or outfit ideas. It works in 45 languages and is free on newer …

Amazon is testing a new AI feature, “Buy for me,” that allows customers to order products from external online stores. Using Nova and Claude AI, the agent selects items and securely enters …

Microsoft has introduced WHAMM (World and Human Action MaskGIT Model), an AI model that generates and allows real-time gameplay of Quake II. It produces over ten frames per second at 640 × 360 pixels …

A new study by Anthropic reveals that AI models often lack transparency in their decision-making processes, even when providing step-by-step explanations. Reasoning models like Claude 3.7 Sonnet and …

[Midjourney(https://www.midjourney.com) has launched an alpha version of its V7 image generation model, offering improved quality and personalization. Users can rate 200 image pairs to customize the …

Microsoft has added new features to its Copilot AI assistant, including Shopping, Pages, and Deep Research. CEO Mustafa Suleyman highlighted Copilot’s goal to be an “AI Companion for …

Meta has launched the first two Llama-4 models, Scout and Maverick, designed to process both text and images. Trained with the Behemoth model, they feature 109 billion and 400 billion parameters, …

Researchers have introduced SeedLM, a new post-training method that compresses large language models by encoding weights as seeds for pseudo-random generators. By reconstructing weights on the fly …

OpenAI is offering ChatGPT Plus for free to all students in the US and Canada until May, just one day after Anthropic launched its “Claude for Education” program for US colleges. The …

Microsoft had announced a $80 billion investment in AI data centers for 2025, but now the company is scaling back several projects. Bloomberg reports delays or cancellations in Indonesia, Australia, …

AMDs Ryzen-AI processors have significant security flaws in their software and drivers, with four vulnerabilities identified. The most critical issues involve the NPU (Neural Processing Unit) driver, …

In just one week, users generated over 700 million images using ChatGPT’s new image feature, with 130 million people engaging with it. Demand is especially high in India, and during the Ghibli-style …

OpenAI’s PaperBench measures how well AI replicates scientific studies. In a test of 20 machine learning papers, human PhD students achieved a 41.4% replication rate, nearly twice as much as the best …

AI systems will be considered medical devices if they make diagnoses, interpret medical data, or assist in therapies. These AI-based medical products will now be regulated by both the MDR and the AI …

AI company Anthropic is entering the education market with a tailored offering for universities. “Claude for Education” features a specialized learning mode that encourages students to …

ChatGPT has seen rapid growth, with paid subscribers increasing by 30% in three months to 20 million and total weekly users reaching 500 million. However, despite these impressive numbers presented …

Meta is developing a new version of its Ray-Ban smart glasses, codenamed “Hypernova,” which will feature an integrated display at the bottom of the lenses. The price is expected to exceed …

Perplexity CEO Aravind Srinivas has dismissed rumors about the company’s financial struggles. While he acknowledged quality issues with the AI’s responses, he assured that these would be …

Google DeepMind is optimistic about achieving artificial general intelligence (AGI) within the next five to ten years, with no significant technical barriers remaining. CEO Demis Hassabis emphasizes …

Wikipedia has seen a significant rise in traffic, stressing its infrastructure. Since January 2024, bandwidth for multimedia content has increased by 50%. This surge is attributed to AI scrapers, …

Meta, in collaboration with the University of Waterloo, has introduced MoCha, an AI system that generates full character animations from text and speech. Unlike previous models, MoCha creates …

Adobe has introduced an update for Premiere Pro with several new AI features. The key addition is Generative Extend, now fully available and supporting 4K footage. This tool extends clips or continues …

OpenAI has launched the OpenAI Academy, a free platform offering dozens of hours of video content and events aimed at providing accessible AI education. The academy includes courses like …

OpenAI plans to release a language model without usage restrictions for the first time since GPT-2. CEO Sam Altman promises a “very, very good model” with advanced logical reasoning …

OpenAI’s upcoming open-weight model marks a massive evolution from GPT-2, transitioning from a 1.5B-parameter text generator to a sophisticated reasoning engine. Unlike GPT-2’s narrow focus on …

Anthropic has developed a method to study the inner workings of Large Language Models (LLMs), revealing unexpected behaviors. The Claude team found that models take contradictory shortcuts to complete …

RunwayML has unveiled Gen-4, a new AI model for video generation that improves dynamic video creation and consistency in styles and objects. A key feature is its ability to generate consistent …

Amazon has introduced Nova Act, an AI agent similar to OpenAI’s Operator, capable of autonomously performing tasks in web browsers. U.S. customers and developers can preview the SDK at …

Investor Peter Thiel played a key role in OpenAI’s internal conflict, warning Sam Altman about tensions between AI safety and commercial interests. The Wall Street Journal reported deeper issues, …

OpenAI has raised a record-breaking $40 billion, with SoftBank contributing $30 billion under the condition that OpenAI fully transitions to a for-profit model by year-end. This funding pushes …

A U.S. judge has allowed The New York Times’ lawsuit against OpenAI to proceed. The newspaper accuses OpenAI of unauthorized use of its articles for AI training and reproducing content in …

OpenAI has adopted a more liberal approach to image generation in ChatGPT, focusing on precise safeguards rather than blanket bans. The goal is to give users more creative freedom while still …

A study from the Hebrew University of Jerusalem found that RAG models perform better with fewer, but highly relevant documents. Tests with Llama-3.1 and Gemma 2 showed up to 10% improvement, as …

German psychologist Luke Bölling successfully exploited psychological manipulation techniques to bypass security measures in various Large Language Models (LLMs). By framing interactions within a …

xAI has acquired X in an all-stock transaction, with the combined companies valued at $80 billion. This merger merges xAI’s rapid growth in AI with X’s extensive user base, creating a platform …

Krisp, a meeting assistant, has added a new feature to its AI-powered desktop app that can filter out accents in real-time during conversations. The initial beta converts Indian accents into standard …

Alibaba released Qwen2.5-VL-32B, an updated multimodal AI model with 32 billion parameters, surpassing similar open models like Gemma 3-27B and Mistral Small 3.1 24B in initial benchmarks. It performs …

Tim Cook praised Chinese AI company DeepSeek’s model during his visit to China. Due to local regulations, Apple is likely to partner with Alibaba’s Qwen model for AI operations in China, …

DeepSeek has released its latest open-source AI model, DeepSeek-V3-0324, under the MIT license. It shows notable improvements in mathematical tasks, outperforming some top models like GPT-4.5 and …

EuroBERT is a multilingual encoder model developed by a research-industry consortium to improve performance in European languages. Built on the BERT architecture, offering faster processing and better …

OpenAI has integrated the Model Context Protocol (MCP) from competitor Anthropic into its products, including the ChatGPT desktop app. MCP standardizes how AI applications access data, allowing …

OpenAI has released a new image generator based on its GPT-4o model, which has quickly gone viral for turning users’ photos into images in the iconic style of Studio Ghibli films like My …

Apple will use images captured by its fleet of camera vehicles and backpacks not only for its Maps service but also for training its AI models and enhancing other products and services.

The ARC-Prize-Foundation has updated its AGI benchmark, ARC-AGI-2, which poses significant challenges to current AI models, with even leading models like OpenAI’s O1 and DeepSeek’s R1 …

eBay has updated its privacy policy to allow the use of personal data for training its AI models, aiming to enhance its services and develop new features. This includes the use of personal information …

xAI has added two new features to its chatbot Grok: an enhanced search function called “DeeperSearch” and an image editor. The “DeeperSearch” function, which replaces the …

Anthropic has improved its AI assistant Claude’s problem-solving abilities with a simple “think” command. This method adds a thinking step, allowing Claude to record its thoughts in …

Google DeepMind has unveiled its latest AI model, Gemini 2.5, which is claimed to be the company’s most intelligent AI to date. The first release, Gemini 2.5 Pro Experimental, leads multiple …

OpenAI has enhanced ChatGPT with a new, more powerful image generator, replacing the previous DALL-E model. The new generator is integrated directly into the Omnimodel GPT-4o, allowing it to process …

Global electricity consumption rose by 4.3% in the past year, driven by the increased use of air conditioning, electric vehicles, and artificial intelligence data centers. This was nearly double the …

Apple has resolved an issue in the macOS 15.4 beta where the storage usage of Apple Intelligence was not displayed. The AI system requires at least 7GB of SSD space, which can be problematic for …

ByteDance has introduced InfiniteYou, an AI system designed to improve portrait photo generation. It uses a new approach with a component called InfuseNet to analyze and extract facial features, …

South Korean chipmaker FuriosaAI turned down an $800 million acquisition offer from Meta, preferring to remain independent and focus on developing AI chips. Meta sought to reduce its dependence on …

Malaysia plans to tighten semiconductor regulations following U.S. pressure to prevent the illegal flow of AI-critical Nvidia chips to China. A special task force will monitor the country’s expanding …

Tencent has introduced the Hunyuan-T1 AI model, which competes with OpenAI’s o1 and DeepSeek’s reasoning models in benchmarks. The model was optimized using extensive reinforcement learning. While it …

Sam Altman announced leadership changes at OpenAI, with a focus on research and product development. Mark Chen is now Chief Research Officer, aiming to drive scientific progress and integrate research …

A joint study by OpenAI and MIT Media Lab examined the emotional impacts of ChatGPT and its Advanced Voice Mode. It found mixed reactions, with heavy users reporting more dependency and worsened …

A British research group has developed an AI system capable of generating complete weather forecasts using just a standard desktop PC. Traditionally, weather predictions required powerful …

The Cybernetic Teammate study by Procter & Gamble with 776 professionals revealed that individuals using AI support can achieve results as good as traditional two-person teams. However, …

Kai-Fu Lee, former head of [Google China (https://en.wikipedia.org/wiki/Google_China), views DeepSeek, a Chinese AI startup offering free open-source models, as an existential threat to OpenAI. …

The European Medicines Agency (EMA) has validated an AI-based diagnostic tool for the first time. The AIM-NASH system by [PathAI] (https://www.pathai.com)assists pathologists in analyzing liver biopsy …

Cloudflare has introduced an AI-generated labyrinth to detect and deter unauthorized bots and AI crawlers from scraping content. The “AI Labyrinth” aims to slow down and confuse these bots …

Perplexity has once again expressed interest in acquiring TikTok, as the platform faces a potential ban in the U.S. Unlike other potential buyers like Oracle, Perplexity is not interested in TikTok’s …