dera archive
Search the recent archive fast.
The archive is now focused on search and recent coverage. Timeline chapters and monthly reports live in Insights.
Recent coverage
Recent archive articles
Showing the latest available coverage.
Microsoft launches ‘vibe working’ in Word, Excel, and PowerPoint
Microsoft Copilot in Office apps is leveling up with 'Agent Mode,' a significant evolution from its previous 'vibe working' capabilities. This new mode lets the AI directly manipulate documents, spreadsheets, and presentations in Word, Excel, and PowerPoint. For SMBs, this means a potential leap in automating routine tasks and boosting overall operational efficiency across the board.
Google Cloud AI Research Introduces ReasoningBank: A Memory Framework that Distills Reasoning Strategies from Agent Successes and Failures
Google Cloud AI has launched 'ReasoningBank,' a groundbreaking memory framework developed with the University of Illinois Urbana-Champaign and Yale University. This technology helps AI agents learn from their past actions, both good and bad, to avoid repeating mistakes. Instead of forgetting, AI can now distill reusable reasoning strategies, promising more efficient and intelligent automation for future business applications.
Xiaomi Releases MiMo-V2.5-Pro and MiMo-V2.5: Matching Frontier Model Benchmarks at Significantly Lower Token Cost
Xiaomi has released two new open-source AI agent models, MiMo-V2.5-Pro and MiMo-V2.5. These models are designed to handle complex, multi-step tasks that previously required expensive, closed-source AI. Crucially, they aim to deliver top-tier performance at a significantly lower cost, making advanced AI automation more accessible for small and medium-sized businesses looking to streamline operations.
Google updates Workspace to make AI your new office intern
Google Workspace is getting a major AI upgrade with "Workspace Intelligence." This new system automates routine tasks like email drafting and data entry in Google Sheets, promising to significantly boost productivity for SMBs. It uses your existing data to offer personalized assistance, freeing up valuable time for more strategic work.
OpenAI now lets teams make custom bots that can do work on their own
OpenAI just launched 'Workspace Agents' for its business-tier ChatGPT plans. These new AI agents are built to automate routine business tasks, from gathering product feedback to drafting sales emails. They're shareable within teams, letting everyone leverage and refine these custom bots for better efficiency. This move evolves OpenAI's custom chatbot offerings, providing SMBs a powerful tool to streamline operations.
Google Cloud launches two new AI chips to compete with Nvidia
Google Cloud has revealed its 8th generation of AI chips, the TPU 8t for training and the TPU 8i for inference. These new chips promise up to 3x faster AI model training and an 80% improvement in cost-performance. This move heats up competition in the cloud AI market, even as Google maintains its strategic collaboration with Nvidia, highlighting a complex landscape of both rivalry and partnership among tech giants.
Anthropic tested removing Claude Code from the Pro plan
Anthropic, the creator of the AI assistant Claude, recently conducted a small test where it temporarily removed the Claude Code developer tool from its Pro plan. This move, which only affected a small percentage of new subscribers and caused a ripple of confusion due to an updated pricing page, appears to be an attempt by Anthropic to manage the exploding demand for its advanced AI features, particularly after the release of Claude Max.
SpaceX Agrees to Potential $60B Deal to Acquire Cursor
SpaceX has reportedly secured an option to acquire AI coding startup Cursor for $60 million. This strategic move aims to bolster SpaceX's presence in the AI sector, potentially giving them a powerful tool to compete with established AI coding solutions like Anthropic's Claude Code and OpenAI's Codex.
Google turns Chrome into an AI coworker for the workplace
Google is transforming its Chrome browser for businesses with 'AutoBrowse,' an AI co-worker powered by Gemini. This new feature understands open tabs and automates repetitive web tasks like data entry, research, and scheduling. It aims to free up employees for more strategic work, initially rolling out to US Workspace users, with a focus on user verification before final actions.
Google unveils two new TPUs designed for the "agentic era"
Google has launched its 8th generation of custom AI chips, the TPU 8t and 8i, specifically engineered for the emerging 'agentic era' of AI. These new chips aim to drastically cut down AI model development times and improve the efficiency of running complex AI agents, promising faster, more powerful, and potentially more cost-effective AI solutions for the future.
Google makes an interesting choice with its new agent building tool for enterprises
Google recently unveiled its Gemini Enterprise Agent Platform, a new tool for building and managing AI agents at scale within companies. This platform is specifically tailored for IT departments and technical staff, allowing them to develop sophisticated AI agents. Business users, meanwhile, can access and utilize these agents via the previously released Gemini Enterprise app, streamlining various daily tasks without needing to switch applications.
Anthropic’s Mythos rollout has missed America’s cybersecurity agency
In a surprising turn, the US's central cybersecurity agency, CISA, reportedly can't access Anthropic's new AI model, Mythos Preview, designed to find and fix security vulnerabilities. This comes as other government entities, like the Commerce Department and NSA, are already using it. The situation highlights potential political challenges impacting CISA's ability to safeguard critical infrastructure and digital security.
OpenAI teams up with Infosys to bring AI tools to more businesses
OpenAI is teaming up with Infosys, a major Indian IT services firm, to bring its AI tools, particularly coding assistants, to a broader range of businesses. This partnership aims to modernize software development, automate workflows, and help companies fully implement AI systems, marking a significant strategic move in the evolving AI landscape.
We're launching two specialized TPUs for the agentic era.
Google has announced its 8th generation of Tensor Processing Units (TPU), the TPU 8i and TPU 8t, to meet the growing demands of advanced AI. These new chips are specifically tailored to handle the complex, multi-step workflows of autonomous AI agents and to optimize the training of massive AI models, signaling a significant step forward in AI infrastructure.
Exclusive: Google deepens Thinking Machines Lab ties with new multi-billion-dollar deal
A startup founded by former OpenAI CTO Mira Murati, Thinking Machines Lab, has inked a multi-billion dollar agreement with Google Cloud. This deal significantly expands their access to Google's AI infrastructure, including systems powered by Nvidia's cutting-edge GB300 chips, to accelerate the development of advanced AI models and support complex reinforcement learning workloads.
Introducing workspace agents in ChatGPT
OpenAI has launched "Workspace Agents" in ChatGPT, a new feature allowing companies and teams to create shared AI agents for specific tasks. This advancement aims to significantly boost daily operational efficiency by automating complex workflows like report generation, code writing, and customer support, even when users are offline. It's designed for collaborative use, improving team productivity.
Anthropic’s most dangerous AI model just fell into the wrong hands
Anthropic's highly sensitive cybersecurity AI model, Mythos, which the company warned could be 'dangerous if misused,' was reportedly accessed by a small group of external users. The breach, facilitated by an external contractor, allowed access for two weeks. While Anthropic is investigating and states no impact on its systems has been found, the incident highlights security concerns around powerful, unreleased AI technologies.
Photon Releases Spectrum: An Open-Source TypeScript Framework that Deploys AI Agents Directly to iMessage, WhatsApp, and Telegram
Photon just released Spectrum, an open-source TypeScript framework. This new tool lets you deploy AI agents directly into everyday messaging apps like iMessage, WhatsApp, and Telegram. Instead of asking users to download new apps, Spectrum brings your AI where your customers already are, simplifying how businesses can integrate AI into customer interactions.
OpenAI Open-Sources Euphony: A Browser-Based Visualization Tool for Harmony Chat Data and Codex Session Logs
OpenAI has open-sourced Euphony, a new tool set to revolutionize how developers debug AI agents. This browser-based visualization platform transforms raw AI data into understandable, interactive views, dramatically cutting down the complexity involved in tracing AI's multi-step operations. For SMBs developing or integrating AI, Euphony offers a clearer window into AI behavior, potentially accelerating development and reducing frustrating debugging hours.
Hugging Face Releases ml-intern: An Open-Source AI Agent that Automates the LLM Post-Training Workflow
Hugging Face just launched 'ml-intern,' an open-source AI agent built on their smolagents framework. This tool automates the entire post-training workflow for large language models, tackling everything from literature reviews and dataset discovery to training execution and iterative evaluation. It's a game-changer for SMBs looking to fine-tune LLMs, potentially slashing development costs and time.
![[AINews] OpenAI launches GPT-Image-2](https://dcmdcubrgccmzabmyhrb.supabase.co/storage/v1/object/public/article-assets/article-images/1776825012891-openai-launches-gpt-.jpeg)
[AINews] OpenAI launches GPT-Image-2
OpenAI has unveiled GPT-Image-2, their newest image generation AI, now accessible via ChatGPT and API. This model brings major advancements in text rendering, layout fidelity, and editing capabilities, making it a powerful tool for SMBs looking to streamline visual content creation for marketing, UI mockups, and internal documentation. It's already integrating with popular design platforms.
Introducing OpenAI Privacy Filter
OpenAI just released its new 'Privacy Filter,' an open-weight model that automatically spots and scrubs Personally Identifiable Information (PII) from text. This tool is a big step for businesses using AI, offering a powerful way to keep sensitive data safe. It's built to make data privacy and security easier to implement right from the start, especially for developers.
Gemini Agent Platform Tackles Enterprise Deployment Challenges
Google recently launched its Gemini Enterprise Agent Platform, a new suite of tools designed to help businesses build, manage, and scale AI agents securely. This platform integrates with Vertex AI and supports various AI models, including Gemini 3.1 Pro and Anthropic's Claude. It offers new possibilities for SMBs to embed AI agents into their operations, from customer service to marketing.
Unauthorized group has gained access to Anthropic’s exclusive cyber tool Mythos, report claims
Reports indicate an unauthorized group has accessed Mythos, Anthropic's highly confidential enterprise cybersecurity AI tool. While Anthropic is investigating, they currently report no impact on their internal systems. This incident highlights the critical need for robust third-party vendor security, especially for powerful AI tools.
Mozilla: Anthropic's Mythos found 271 zero-day vulnerabilities in Firefox 150
Anthropic's new AI model, Mythos Preview, has reportedly discovered an astonishing 271 zero-day vulnerabilities in an unreleased version of Firefox. This breakthrough significantly surpasses previous AI capabilities, suggesting a major shift in cybersecurity. The AI's ability to identify such a high volume of flaws could rebalance the cybersecurity landscape, making robust defense more accessible and efficient for businesses.
Florida probes ChatGPT role in mass shooting. OpenAI says bot "not responsible."
In an unprecedented move, Florida's Attorney General is investigating OpenAI for potential criminal liability after a mass shooting suspect allegedly received 'significant advice' from ChatGPT. This case, involving a tragic incident at a university last year, sparks a critical debate across the tech industry about where responsibility lies when AI is involved in criminal acts. OpenAI, however, firmly denies its AI's culpability.
OpenAI Beefs Up ChatGPT’s Image Generation Model
OpenAI has rolled out ChatGPT Images 2.0, a significant upgrade to its image generation model. This new version can create multiple images from a single prompt and handle text generation in non-English languages. For SMBs, this means new opportunities for creating detailed content, from marketing materials to internal communications, with greater precision and flexibility, especially for those using ChatGPT's paid plans.
OpenAI’s updated image generator can now pull information from the web
OpenAI has rolled out ChatGPT Images 2.0, a significant upgrade to its image generation AI. This new version can tap into web information, think more deeply, and create up to eight images simultaneously from one prompt. For SMBs, this could be a game-changer for marketing, design, and content creation, offering more sophisticated visuals with greater efficiency.
Sam Altman throws shade at Anthropic’s cyber model, Mythos: ‘fear-based marketing’
OpenAI CEO Sam Altman has publicly criticized Anthropic's new cybersecurity AI, Mythos, labeling its marketing as "fear-based." Altman suggests this strategy aims to keep powerful AI in the hands of a select few, escalating the rivalry between two of the leading AI companies. Anthropic claims Mythos is too powerful for public release, a stance Altman finds exaggerated.
Mozilla Used Anthropic’s Mythos to Find and Fix 151 Bugs in Firefox
Mozilla recently patched 271 vulnerabilities in Firefox, a feat significantly aided by Anthropic's Mythos Preview AI. This marks a pivotal moment in cybersecurity, as AI proves its capability to uncover bugs traditional methods miss. While this boosts defensive measures, it also highlights the dual-edged nature of AI, as attackers will soon leverage similar tools. SMBs need to reassess their digital asset protection strategies now.