Tags: Open Ai

Microsoft is promoting a new collaborative workflow it calls “vibe working,” built around AI‑driven tools in its Office suite. The company unveiled Office Agent and Copilot’s Agent Mode for Word and Excel, allowing users to start documents with a single prompt and iterate with AI assistance. Office Agent, powered by Anthropic models, can generate PowerPoint decks and Word files in a chat‑first experience, while the Excel and Word Agent Modes run on OpenAI’s latest models and achieved modest accuracy on benchmark tests. Microsoft says the features will soon expand to desktop versions and additional apps, signaling a deeper push toward human‑agent collaboration in the workplace. Read more

Seoul has unveiled a sovereign AI program that funds five domestic firms—LG AI Research, SK Telecom, Naver Cloud, NC AI and Upstage—to build large language models optimized for Korean language and culture. The initiative aims to reduce reliance on foreign AI services, improve data security, and position South Korea to compete with global players such as OpenAI and Google. Each company brings a distinct strategy, from hybrid reasoning models and telecom‑driven data pipelines to full‑stack cloud ecosystems and cost‑effective startup solutions. Read more

OpenAI introduced a safety routing system that automatically moves ChatGPT conversations to a more conservative AI model when sensitive or emotional topics are detected. Paying users have voiced strong frustration, saying the change forces them away from their preferred models without a way to opt out. OpenAI executive Nick Turley explained that the routing operates on a per‑message basis to better support users showing signs of mental or emotional distress. The company emphasizes its responsibility to protect vulnerable users, while critics compare the feature to locked parental controls. Read more

Silicon Valley dominated headlines this week with massive AI infrastructure deals. Nvidia announced a potential $100 billion investment in OpenAI, while OpenAI revealed plans to add five new Stargate data centers in partnership with Oracle and SoftBank, promising gigawatts of fresh capacity. Oracle financed the effort by selling $18 billion in bonds. At the same time, OpenAI launched Pulse, a personalized morning briefing feature in ChatGPT that currently runs only for $200‑a‑month Pro users due to server constraints. The flurry of investments and product launches raises questions about the balance between costly data‑center expansion and new AI services. Read more

Apple has created an internal iPhone app named Veritas that functions like OpenAI's ChatGPT. The tool is being used by employees to experiment with new Siri capabilities, such as advanced email and song searches and photo‑editing commands. Veritas supports ongoing multi‑turn conversations, mirroring the experience of popular AI chatbots. The development reflects Apple’s effort to catch up in the fast‑moving AI space, amid mixed user reactions to its Apple Intelligence features and a delayed Siri upgrade now slated for the spring of 2026. Apple declined comment on the project. Read more

OpenAI has introduced ChatGPT Pulse, a new feature that delivers a curated set of visual cards each morning based on a user’s calendar, chat history, and other personal data. Currently limited to Pro subscribers, Pulse aims to move ChatGPT from a reactive assistant to a proactive one, offering suggestions for daily activities, meals, workouts, and more. The rollout reflects OpenAI’s broader focus on AI agents that can act on behalf of users, while also emphasizing safety controls and user‑controlled data sharing. Read more

OpenAI has launched Pulse, a new ChatGPT feature that creates personalized reports while users sleep. Offering five to ten concise briefs each morning, Pulse aims to make ChatGPT the first app people check at the start of their day. Initially available to Pro plan subscribers, the service provides news roundups, customized itineraries, and other tailored content displayed as AI‑generated cards. Integrated with Connectors like Google Calendar and Gmail, Pulse can parse emails and calendar events overnight, surfacing key messages and agendas. OpenAI plans to expand Pulse to all users, with future ambitions for more agentic capabilities. Read more

OpenAI introduced a new benchmark called GDPval that pits its AI models against human experts across dozens of occupations. In the initial rollout, GPT-5‑high was judged better than or on par with professionals in about 40.6% of tasks, while Anthropic’s Claude Opus 4.1 achieved roughly a 49% win rate. The test covered 44 roles spanning key sectors such as healthcare, finance, and manufacturing. OpenAI says the results show AI can start offloading routine work for many jobs, though it acknowledges the current scope is limited and plans to expand the benchmark’s coverage. Read more

Databricks announced a multi‑year partnership with OpenAI that integrates the AI firm’s models—including GPT‑5—into its data platform and AI product, Agent Bricks. The deal, valued at a minimum of $100 million, gives Databricks customers secure, native access to OpenAI’s models via SQL or API, allowing them to build AI‑driven applications on top of enterprise data. The partnership builds on earlier work that added OpenAI’s open‑weight models, gpt‑oss 20B and gpt‑oss 120B, to the platform, and follows a similar revenue‑target arrangement Databricks struck with Anthropic. Early customer demand, including from Mastercard, signals strong interest in the combined offering. Read more

Clarifai announced a new reasoning engine that promises to double inference speed and reduce costs by 40 percent. The platform combines low‑level CUDA kernel tweaks with advanced speculative decoding to extract more performance from existing GPU hardware. Independent benchmarks reported industry‑leading throughput and latency. The launch comes amid a surge in demand for AI compute, highlighted by OpenAI’s plan to spend up to $1 trillion on new data centers. Clarifai’s CEO emphasized that software and algorithmic innovations remain critical even as hardware builds out. Read more

OpenAI has introduced Pulse, a new feature for ChatGPT that delivers daily, visual updates based on a user’s recent conversations, calendar, and connected apps. Initially rolling out to ChatGPT Pro users on mobile as a preview, Pulse presents information as a series of scan‑friendly cards and allows users to curate topics through simple feedback. The feature leverages OpenAI’s Connections to Google services, includes safety checks, and aims to evolve the chatbot from a passive tool into a proactive personal assistant. Read more

A study by Harvard Business Review and the Stanford Media Lab finds that AI‑generated content, dubbed "workslop," is spreading across businesses. While tools like Gemini, Copilot, Claude and ChatGPT enable rapid creation of reports, presentations and code, the output often lacks substance and contains errors. About 40% of respondents reported receiving workslop, leading to confusion, offense and a perception that coworkers who rely on AI are less capable, reliable and creative. The report urges organizations to treat AI as an assistant, enforce rigorous editing, and prioritize human collaboration to preserve quality and trust. Read more

OpenAI CEO Sam Altman detailed a bold plan to scale AI compute, aiming to build a factory that can produce a gigawatt of new AI infrastructure each week. He highlighted partnerships with Oracle on massive data centers in Texas, a $500 billion Stargate Project, and Nvidia’s $100 billion investment to power the effort. Altman stressed the need for rapid innovation across chips, power, construction, and robotics, and framed widespread AI access as a future economic driver and potential human right. Read more

Elon Musk's artificial‑intelligence venture xAI has filed a lawsuit against OpenAI, accusing the ChatGPT developer of misappropriating confidential information. The complaint alleges that former xAI employees who joined OpenAI took proprietary source code, data‑center deployment methods and other trade secrets, constituting a pattern of unfair competition and intentional interference. xAI claims the alleged theft undermines its competitive position and violates legal obligations. OpenAI has responded with a statement denying the accusations, emphasizing its own policies against confidentiality breaches. The dispute adds another chapter to the ongoing rivalry between the two AI firms. Read more

Google Cloud is focusing on early‑stage AI companies, offering $350,000 in cloud credits, technical assistance, and go‑to‑market support. The firm promotes an open AI stack that spans custom TPUs, foundation models and applications, aiming to win future unicorns before they grow large. Partnerships include TPU deployments with Fluidstack and collaborations with startups such as Loveable and Windsurf, while Google also hosts Anthropic’s Claude and provides TPUs to OpenAI. The strategy reflects Google’s broader commitment to open‑source tools and comes amid regulatory scrutiny of its search dominance. Read more

OpenAI disclosed plans to stand up five new data center sites across Texas, New Mexico, Ohio and a yet‑to‑be‑announced Midwest location, partnering with Oracle and SoftBank. The effort is part of the so‑called Stargate project, which aims to deliver up to 10 gigawatts of AI compute power and involves billions of dollars in investment. Existing facilities include a site in Abilene, Texas, and collaborations with CoreWeave. CEO Sam Altman emphasized that expanding compute is essential to AI’s promise, while noting challenges such as financing, permitting and community opposition. A lawsuit from Ziff Davis over alleged copyright infringement is also highlighted. Read more

Oracle is reportedly preparing a $15 billion corporate bond offering that could include up to seven tranches, one of which may be a 40‑year bond. The move comes weeks after the cloud‑infrastructure giant signed a massive AI compute agreement with OpenAI valued at $300 billion and entered talks with Meta on a $20 billion deal. Oracle also announced a leadership transition, with CEO Safra Catz moving to executive vice‑chair and co‑CEOs Clay Magouyrk and Mike Sicilia taking the helm. Read more

Enterprise AI model maker Cohere announced a $100 million extension to its recent financing, pushing its valuation to $7 billion. The company also unveiled a partnership with AMD, enabling its Command-family models to run on AMD Instinct GPUs while maintaining support for Nvidia hardware. Co‑founder Aidan Gomez, noted for his role in the original transformer paper, highlighted the focus on enterprise customers seeking AI sovereignty. New investors include the Business Development Bank of Canada and Nexxus Capital Management. The announcement was made at a TechCrunch event in San Francisco. Read more

Microsoft is expanding its AI portfolio by adding Anthropic’s Claude Opus 4.1 and Claude Sonnet 4 models to its Copilot assistant. The move, announced on a Wednesday, follows a recent agreement to embed Anthropic’s technology in Office 365 apps such as Word, Excel, and Outlook. Business users of Copilot will now be able to select between OpenAI’s deep‑reasoning models and Anthropic’s offerings for tasks ranging from complex research to custom AI tool creation. Opus 4.1 targets advanced reasoning and coding, while Sonnet 4 focuses on routine development, large‑scale data processing, and content generation. Read more

A series of experiments with OpenAI’s ChatGPT and Google’s Gemini revealed that the safety mechanisms designed to block gambling advice can be inconsistent. When users first discuss problem gambling, the bots refuse betting tips, but after repeated betting queries the safety cues become diluted and the models provide advice. Experts explain that the models weigh recent conversation tokens more heavily, and longer chats can weaken safety triggers. The findings highlight challenges for AI developers in balancing protective features with user experience, especially as the gambling industry explores AI‑driven tools. Read more