# AI News Recap — October 2025 ## Quote of the Month *“The next year of AI products will establish user habits for many years. People are building their new habits right now, like how quickly everyone started relying on ChatGPT. This creates incredible urgency because whoever captures these habits now will have a lasting advantage. Google recognized they couldn’t let users develop the habit of going elsewhere for AI-powered answers. This is the critical window for establishing how people will search and find information for the next decade.” Robby Stein, VP of Product at Google* ## Microsoft - **[Windows 11 AI PCs](https://blogs.windows.com/windowsexperience/2025/10/16/making-every-windows-11-pc-an-ai-pc/)**: Windows 11 now puts Copilot at the center of every PC, letting you use AI voice commands, screen analysis, task automation, and deep integration with files, apps, and cloud services, including Google tools. - **[Copilot’s Human-centered AI release](https://www.microsoft.com/en-us/microsoft-copilot/blog/2025/10/23/human-centered-ai/)** with 12 new features, including: - **Groups**: Real-time collaborative sessions for up to 32 people. - **Imagine**: Community-driven AI creation remixing. - **Mico**: Customizable Copilot character for warmer interactions. - **Real Talk**: Conversation modes for authentic, challenging discussions. - **Memory & Personalization**: Long-term user preference recall with full user control. - **Connectors**: Integrations with OneDrive, Outlook, Gmail, Google Drive, and Calendar. - **Proactive Actions**: AI-suggested next steps based on recent work. - **[Copilot Mode in Microsoft Edge](https://www.microsoft.com/en-us/edge/copilot-mode?form=MG0AWI&cs=1873324239)** brings powerful AI directly to your browser, letting you search, chat, navigate, summarize content, and control your experience by voice. - New Copilot features - **[Agent Mode and Office Agent](https://www.microsoft.com/en-us/microsoft-365/blog/2025/09/29/vibe-working-introducing-agent-mode-and-office-agent-in-microsoft-365-copilot/)**. Agent Mode lets Copilot act as an autonomous, reasoning AI agent within apps like Excel and Word to orchestrate multi-step workflows and document creation. Office Agent generates documents directly from chats. - **[Microsoft 365 Copilot App Builder and Workflows](https://www.microsoft.com/en-us/microsoft-365/blog/2025/10/28/microsoft-365-copilot-now-enables-you-to-build-apps-and-workflows/)**: The new vibe coding tool enables you to build applications and automate workflows with conversational prompts. - Microsoft also introduced the **[Agent Framework](https://azure.microsoft.com/en-us/blog/introducing-microsoft-agent-framework/)** a developer-focused, open-source toolkit and runtime for building, orchestrating, and managing advanced, multi-agent AI systems. - **[Researcher with Computer Use in Copilot](https://techcommunity.microsoft.com/blog/microsoft365copilotblog/introducing-researcher-with-computer-use-in-microsoft-365-copilot/4464766)**: It's an AI-powered tool that autonomously researches, interacts with web content, and performs actions on your behalf using a secure virtual computer, enabling deeper insights and comprehensive reports by combining public, gated, and work data. - **[Copilot for Health](https://www.microsoft.com/en-us/microsoft-copilot/for-individuals/do-more-with-ai/ai-for-daily-life/copilot-for-health-your-ai-companion-for-smarter-medical-decisions?form=MA13KP)** is Microsoft's AI companion that helps users find credible medical information, understand complex health topics, and locate healthcare providers. - **[Podcasts in Microsoft Copilot](https://support.microsoft.com/en-us/topic/using-podcasts-in-microsoft-copilot-11c7e451-e810-4a82-b619-f96d9badc0a8)**: Copilot Podcasts is a new way to create personalized, on-demand audio content on topics that interest you. - **[MAI-Image-1](https://microsoft.ai/news/introducing-mai-image-1-debuting-in-the-top-10-on-lmarena/)** a new text-to-image model that debuted in the top 10 on LMArena for speed and photorealistic quality. ## Google - **[Gemini Enterprise](https://cloud.google.com/blog/products/ai-machine-learning/introducing-gemini-enterprise):** Google introduced Gemini Enterprise, a full-stack AI platform for the workplace featuring smart agents, workflow automation, and seamless data integration from platforms like Salesforce, SAP and Google Workspace. - **[AI Studio Vibe Coding](https://blog.google/technology/developers/introducing-vibe-coding-in-google-ai-studio/)**: Now lets users build apps by simply describing them in everyday language inside Google AI Studio. - **[OPAL](https://opal.withgoogle.com/landing/):** Opal is an experimental Google tool that lets you build, edit, and share AI-powered mini-apps simply by using natural language, no coding required. - **[Veo 3.1](https://developers.googleblog.com/en/introducing-veo-3-1-and-new-creative-capabilities-in-the-gemini-api/):** Google released Veo 3.1, an update to its AI video model, adding synced sound, improved editing tools, and creator controls. - **[Google Skills](https://blog.google/outreach-initiatives/education/google-skills/):** Launched a worldwide platform with over 3,000 AI-powered courses for upskilling, job search, and certifications across industries. - **[DeepMind introduces CodeMender](https://deepmind.google/discover/blog/introducing-codemender-an-ai-agent-for-code-security/):** An AI agent for automatic code security fixes, patching vulnerabilities at scale. - **[Gemini Robotics 1.5](https://deepmind.google/discover/blog/gemini-robotics-15-brings-ai-agents-into-the-physical-world/)** enables AI agents to interact with and control physical robots, bridging the gap between digital intelligence and real-world actions. - **[Google Pomeli](https://labs.google.com/pomelli/about/)** is an AI-powered marketing tool from Google Labs that enables businesses to quickly create brand-consistent images, product visuals, and promotional content with customizable templates and guided workflows. - **[Workspace Flows](https://sites.google.com/view/workspace-flows/about)** is a new tool that lets you automate tasks and connect Google Workspace apps together with customizable workflows. - **[Google's first AI-generated TV ad](https://9to5google.com/2025/10/31/google-ai-tv-ad/)**: Google's first AI-generated TV ad showcases how AI tools can produce creative and engaging commercials for brand storytelling. - [**Gemini 2.5 "Computer Use" Model**](https://blog.google/technology/google-deepmind/gemini-computer-use-model/): Enables agents to directly interact with web and mobile interfaces like clicking, typing, and filling forms to automate tasks through visual reasoning at high speed and accuracy. - **[Virtual Try-On](https://support.google.com/merchants/answer/16159685?hl=en):** Google expanded its AI-powered virtual try-on, letting users see shoes and apparel on their own images before buying. - **[NotebookLM Mobile: Source Discovery](https://www.androidauthority.com/notebooklm-google-drive-discover-sources-3605246/):** Source Discovery is a new feature in NotebookLM that lets users search for and add Google Drive documents as sources. - **[Google Earth AI](https://blog.google/technology/research/new-updates-and-more-access-to-google-earth-ai/):** Google Earth AI received new features and expanded public access, enabling advanced mapping, environmental research, and geospatial analysis with powerful AI tools. ## OpenAI - **[DevDay 2025 Announcements](https://openai.com/devday/):** At its annual developer event, OpenAI unveiled groundbreaking features including: - [The GPT-5 Pro API ](https://platform.openai.com/docs/models/gpt-5-pro)for advanced reasoning tasks. - [Apps inside ChatCPT - The Apps SDK](https://openai.com/index/introducing-apps-in-chatgpt/) that lets developers create full interactive apps directly inside ChatGPT. There will be third party services available like Booking, Canva, Zillow, and others. - The [AgentKit](https://openai.com/index/introducing-agentkit/) toolkit for building advanced AI agents and workflows more rapidly. - **[Sora 2 Video Generation](https://openai.com/index/sora-2/):** Sora 2, the next-gen AI video model from OpenAI, now generates lifelike video with synchronized audio, and introduces a character cameo feature for integrating custom avatars and pets into generated scenes. - **[Aardvark – Automated Security](https://openai.com/index/introducing-aardvark/):** OpenAI debuted Aardvark, a GPT-5-powered agent that autonomously finds, validates, and fixes code vulnerabilities at scale, bringing stronger security to enterprise development workflows. - **[ChatGPT Atlas Browser](https://chatgpt.com/atlas):** OpenAI launched ChatGPT Atlas, a browser app that brings ChatGPT’s AI assistant to any website, providing instant insights, explanations, and suggestions while you browse. Currently only for macOS. - **[Codex Upgrades and IDE Extension](https://developers.openai.com/codex/ide):** The Codex IDE extension lets you use OpenAI's coding agent to write and automate code directly in your editor with chat and smart prompts. - **[Adult Conversation Support](https://elink471.genai.works/ss/c/u001.ykupDz589Rh8qoraylLTIUE5of05cM4kcFjLnqNrCG8SRpx9dzuuNu70ZCjXcMltRrlMQQmFtvimy1DIZyWwoRe6-Wapg32qA5dM0bGpCKgk1kX9EePH72qL1C9JrnOSEewNIJz3V2LLbXPf9kOtDENIto3WqZkHrEANOgKFvDpnThCiKVoh-ykLcnAIvHERNZOgCXhQ6lt3xumQetCvzXwC4vP0RRsfBVa_XcOZRCfdldWeb7QBkc35LMKhEKf2fs80BOp6-WXXW58Y0JsAw8_tlupwBPZIrQFpriJrML0T7NrCLAmkHZxAd5NrRla5/4kr/ftXZvZvGR4qcglFunoGWVg/h7/h001.gg-F5OjxrmFmp_gtgyP1M0qx6YmAOIb7PgBlrf2H17I):** CEO Sam Altman confirmed that ChatGPT will soon introduce verified adult erotic conversation capabilities, a move aimed at making chatbot interactions sound more human and natural. - **[OpenAI Transitions to For-Profit](https://openai.com/index/next-chapter-of-microsoft-openai-partnership/):** OpenAI announced it has become a for-profit company, signaling a new era of scaled AI productization and commercialization. That also means redefined partnership with Microsoft. - **[PayPal x OpenAI Partnership](https://newsroom.paypal-corp.com/2025-10-28-OpenAI-and-PayPal-Team-Up-to-Power-Instant-Checkout-and-Agentic-Commerce-in-ChatGPT):** PayPal finalized a deal with OpenAI to embed its digital wallet into ChatGPT, paving the way for seamless shopping experiences directly from conversations. ## Anthropic - **[Claude Haiku 4.5](https://www.anthropic.com/news/claude-haiku-4-5):** Claude Haiku 4.5 is Anthropic’s newest lightweight AI model designed for fast, efficient, and reliable natural language tasks. - **[Claude for Life Sciences](https://elink471.genai.works/ss/c/u001.2HFabJfhvL4m05haQHN-pyvXS6xkhRoU1n-avOA8utp0J1OHWfDF9fIaoDSoplbcUraROsSG3crueSsfyNdU38doxngMdEgm-MmMqhx_wcis8lzFFSSjATFuoGpybbWL6-eLr4CxB8wu1RMh5JfuStng0ABUrkbU_4qJACD2j9Ve97rREiMozsMOG2xroarvbZqUoMRUdjS6ryuo8Jbosge7MEaMp9mdu3B9DTWNUU24kIx01wBGngiXk7ZK9QUuxJaEHhZVdSWnLA8Cl5PljHP4nJNzKKyvvpeY68XiwccavjV0QEEMXqvWbouak4CS/4kx/6cm8APsnQxaqj-Yi6bgE7A/h7/h001.xP8IMF60wOP_iToiGCkzLIs2av7zjI9QyfVPqjASD5o):** New industry-focused platform designed to help biomedical researchers accelerate discoveries and apply AI to real-world medical problems more efficiently. - **[Claude Skills for Custom Workflows](https://www.anthropic.com/news/skills):** Anthropic unveiled Claude "Skills," which are customizable folders of instructions and resources that enable Claude to perform advanced business and brand-specific workflows on demand. - **[Claude Code Plugins](https://www.anthropic.com/news/claude-code-plugins):** Claude Code now supports web-based plugins, letting users add specialized tools and automation capabilities for truly extensible AI development environments. - **[Claude Code on the Web](https://www.anthropic.com/news/claude-code-on-the-web):** A new browser-based code assistant that helps you write, debug, and review code directly in your web browser. - **[Petri Open-Source Auditing](https://www.anthropic.com/research/petri-open-source-auditing):** A tool for automated safety audits of large language models, helping the industry advance responsible and secure AI. ## Adobe - **[Adobe Creative Cloud AI Updates](https://news.adobe.com/news/2025/10/adobe-max-2025-creative-cloud):** Adobe delivered major AI innovations across its suite (Photoshop, Lightroom, Premiere, Illustrator, and others) now all integrate AI co-pilots for editing images, video, and design via natural language prompts. The Cloud AI update includes: - **Firefly Image Model 5:** The new model delivers ultra-photorealistic and 4MP resolution images, advanced photorealism, accurate human anatomy generation, and direct text-in-image editing from conversational or written commands. - **AI Marketplace and Model Customization:** Creative Cloud now offers centralized access to Firefly, OpenAI, Google Imagen and Veo, Runway, Pika, Topaz, and more, and lets users train bespoke models with as few as 10 images for brand-charged results. - **Conversational Editing Assistants:** New chat-based co-pilots in Photoshop and Express help automate advanced editing tasks, offer design ideas, and execute complex sequences in one step. - **Premiere Video & Audio AI:** Premiere now supports AI-powered video generation from text, automated subject tracking, transcript audio editing, and one-tap publishing to YouTube Shorts. - **Enhanced Lightroom Tools:** AI now helps sort, cull, and retouch bulk photo collections, removing dust, spots, or unwanted reflections instantly. ## Other Notable Updates - **[DeepSeek OCR](https://deepseek.ai/blog/deepseek-ocr-context-compression):** DeepSeek released an open-source OCR system able to process 200,000 pages daily on one GPU, setting new efficiency benchmarks for optical character recognition. - **[Amazon Quick Suite](https://www.aboutamazon.com/news/aws/amazon-quick-suite-agentic-ai-aws-work):** Amazon launched Quick Suite, a next-gen agentic platform for automating workflows and research orchestration across cloud environments. - **[Slack AI Upgrades](https://slack.com/features/ai):** Slack rolled out new AI bot features including project plan drafting, priority automation, and a ChatGPT-powered team app, streamlining office collaboration. - **[NVIDIA DGX Spark](https://tsttechnology.io/blog/ai-tech-trends-oct-2025):** NVIDIA began shipping DGX Spark, a compact AI supercomputer capable of delivering a petaflop of compute power for local model training and deployment. - **[Apple M5 Chip](https://elink471.genai.works/ss/c/u001.2HFabJfhvL4m05haQHN-pwV3fMWokz4t7v6mU96-68mmo9w4wOQybrq_pYdfMcPCtRTb8DKXtaw_JgxfAC7PFjfMAcO24cXqLdtfFUFzNxfZWHI3IOm3-SzT7F6yUZdVYpwhBKistC4lTvyIf4kEmuSi435dbZogiqFBIj3v9FZ3CQSebHaXYO9IMLS7K4Uc9om-eWAHL1Tv1l0BLJ7iwC-fM6E6Vt7vx36KQ4lkgvQUHlQ5eIPL2Miw6wTFvHcA0vsTiLejUO-bj6g9mdzRUvobhG4cdiNw7bR9qgi621oNToon7g0EM3tDLC-533DsFquo55qqdSkJG_qRo4OlejTBKU4UTXToUmH1yOq86g53R4ULKVGr3_GNrOTz47wI/4kt/-tZNm11RT_a5zkWcyzXWEg/h8/h001.OSLgXvC-rGXWEyrGiUP4BOCg8_WZHSYan-tfKOGRp1w):** Apple introduced its M5 chip, featuring neural accelerators embedded in every GPU core, quadrupling the AI computing power of the M4 generation and supporting next-level on-device AI workloads. - **[Perplexity Patents Search](https://www.perplexity.ai/hub/blog/introducing-perplexity-patents):** Perplexity debuted Patents, a free AI-driven patent, IP, and paper search engine powered by natural language queries. - **[GitHub Agent HQ](https://link.mail.beehiiv.com/ss/c/u001.iqGpp1oGBQ0uktaT8DrDbt50fq5jSoMohdcsNDag6nHZHGqeZSjHeCqK6D4TNx5-TCMnZM7p-4Idwbk70md5bvf0aDuBFwXcTLYkjUUpJ4PNrTZffx1UEueohubJK18tNA-c-3LX9UIgoVfItG-Ag35b-6Xq9VJY2q_QIlXaYPpFwIqiZSUdZ53d8ddDYqvIHwGzBeDv0CyQlwApWezOYLrnQjS6E0S9l9ed-6kz81LCW9d5IQTmO-jtoUT27p5wYCMZUYItqJBYwlfUDh8jsg/4l9/VNIeMdWgT_WE-HQWoK_aHg/h37/h001.7Ue0kVnx3VG1LuXHZ9aeutZh2KxjQdnNhgd1h2gjjy8):** GitHub released Agent HQ, a platform for seamlessly integrating agents from Anthropic, OpenAI, Google, and others into existing developer workflows. - **[Mistral Studio/Le Chat Updates](https://link.mail.beehiiv.com/ss/c/u001.d5-nKkYi8xJQ2TdkbjBQyqPFdLvPeoZSY9VZZRScz4I-6AT4gzIF-ygqQvlxm53BqQdYZvcSRs7DJ6DoiXkFptiQpomJuDFpnc5hkf2D67oNvExuywtIVTTAkdPjDCDSoV8W0bbyzbdQuBGQbsIrk-w9WN7k4_KGemRnzADBIjEt5ySDraB3IgfhUCnO1oRuZamVZoi-RhlhwzAY7uVNMw/4l9/VNIeMdWgT_WE-HQWoK_aHg/h38/h001.DIF1QpJiS7X3336TrkUxn1ycMxxXlqr-4WwAfD0uaY8):** Mistral AI improved its Studio for scalable agent deployment and rolled out "Memories" and connectors to Le Chat for smarter automated workflows. - **[Grokipedia by X](https://grokipedia.com/):** Elon Musk launched Grokipedia, a straightforward online AI-powered encyclopedia for accessible knowledge sharing. - **[HuggingChat Omni](https://huggingface.co/chat/)**: A policy-based router that dynamically selects among approx. 115 open-source LLMs per prompt. - **[ElevenLabs Visual Voice Agent Builder](https://elevenlabs.io/docs/agents-platform/customization/agent-workflows)** (“Workflows”): A drag-and-drop graph editor for branching, real-time voice agents, with sub-agents, toolcalls, and testing harness. - **[Karpathy’s nanochat](https://github.com/karpathy/nanochat)**: An open-source, minimal, full-stack implementation for training and running your own large language model, designed to be hackable, fully owned, and run end-to-end by individuals on a single GPU node. - Thinking Machines Lab - **[Tinker API](https://lnkd.in/gQBS2CcQ)**: A flexible API and managed service for fine-tuning open-weight language models and designed to make model customization accessible to researchers and developers. - **[Canva Creative Operating System](https://www.canva.com/newsroom/news/creative-operating-system/)**: Unifies design, collaboration, brand ops, and AI into a single workspace. - **[Manus 1.5](https://manus.im/blog/manus-1.5-release)** delivers faster, more reliable, and higher-quality performance across tasks, including full-stack web app development, and introduces advanced collaboration features, an expanded context window, new agent modes, and a centralized library for generated files. - **[Cursor 2.0 and Composer](https://cursor.com/blog/2-0)**: A fast new coding model, and a multi-agent interface that lets you run many agents in parallel, making coding and code review more efficient. - **[The X1 Neo home robot](https://www.1x.tech/)** is available for pre-order, but it does not yet perform expected household tasks and is intended for early adopters to help train and improve its capabilities.