Anthropic moves beyond the chatbox — into your computer.
Meta plans $10B cable project, Alibaba, Ai2, Stability make more open source moves, Perplexity mulls getting into hardware.
Foundational models are breaking out of the chat box — and into your computer.
This week, Anthropic released Model Context Protocol (MCP) — something like a universal connector to pipe data and tools directly to Claude. This follows moves from OpenAI’s Work With apps announced earlier this month, which lets ChatGPT connect directly to coding apps, like Xcode.
When foundational models can directly interact with our tools and data, everything changes.
Early observations:
Dynamic apps — personalized, ephemeral, real-time — become dramatically easier to build.
This could usher in the next phase of agentic workflows where the model creates AI agents independently without input from the human.
While these releases are aimed primarily at developers for now, the next era of computing will fundamentally change the end user experience in technology.
Over the next few weeks, we will be diving deeper into the next era of human-machine experiences, as we head into 2025.
Happy holidays!
The Latest This Week
Meta plans to build a $10B subsea cable spanning the world, sources say. Meta, the parent of Facebook, Instagram, and WhatsApp, is the second-biggest driver of internet usage globally, accounting for 10% of all fixed and 22% of all mobile traffic. Meta will be the sole owner and user of this subsea cable.
OpenAI’s Sora video generator appears to have leaked: A group appears to have leaked access to Sora, OpenAI’s video generator, in protest of what it’s calling duplicity and “art washing” on OpenAI’s part. The group published a project on the AI dev platform Hugging Face seemingly connected to OpenAI’s Sora API, which isn’t yet publicly available.
Alibaba releases an ‘open’ challenger to OpenAI’s o1 reasoning model: A new so-called “reasoning” AI model, QwQ-32B-Preview, has arrived on the scene. It’s one of the few to rival OpenAI’s o1, and it’s the first available to download under a permissive license. Developed by Alibaba’s Qwen team, QwQ-32B-Preview contains 32.5 billion parameters and can consider prompts up ~32,000 words in length; it performs better on certain benchmarks than o1-preview and o1-mini, the two reasoning models that OpenAI has released so far.
Anthropic launches tool to connect AI systems directly to datasets: Anthropic has released a new open-source tool to connect AI assistants directly to the information they need to inform their responses or carry out tasks. The new Model Context Protocol (MCP) provides a universal connection to all sorts of data sources, which Anthropic says will improve performance.Anthropic says coding software like Replit and Codeium have already started using MCP to build out their AI agents, which can complete tasks on behalf of users.
Ai2 releases new language models competitive with Meta’s Llama: Ai2, the nonprofit AI research organization founded by the late Microsoft co-founder Paul Allen, released OLMo 2, the second family of models in its OLMo series. Ai2 claims the result is models that are competitive, performance-wise, with open models like Meta’s Llama 3.1 release.
Luma expands Dream Machine AI video model into full creative platform, mobile app: Luma AI, a startup founded by former Googlers and others, is dramatically expanding its Dream Machine AI video model with a new interface, mobile app and new image generation foundation model. The model, Luma Photon, combines personalization, efficiency and creative power to push the boundaries of image and video creation.
ElevenLabs’ new feature is a NotebookLM competitor for creating GenAI podcasts: ElevenLabs introduced a feature that lets users upload different types of content to create a multispeaker podcast for them, similar to Google’s NotebookLM. The feature, called GenFM, can be found on the company’s ElevenLabs Reader iOS app. Currently, GenFM supports 32 languages, including English, Hindi, Portuguese, Chinese, Spanish, French, German, Japanese, and others.
Anthropic says Claude AI can match your unique writing style: Anthropic is adding a new feature to its Claude AI assistant that will give users more control over how the chatbot responds to different writing tasks. This update aims to personalize the chatbot’s replies and make them feel more natural or appropriate for specific applications, such as writing detailed technical documents or professional email.
Amazon invests additional $4 billion in Anthropic: Amazon has made another big investment in foundation model provider Anthropic. Amazon revealed that it is now Anthropic's primary AI model training partner. Anthropic will also use AWS Trainium and Inferentia AI chips to train and deploy its future foundation models. The new investment brings Amazon's total investment in the AI startup to $8 billion.
Perplexity mulls getting into hardware: Perplexity, the AI-powered search engine, wants to get into hardware — kinda sorta. Aravind Srinivas, Perplexity’s founder and CEO, posted on X that he was considering making a “simple, under $50” device to “reliably answer” questions “voice to voice.” He promised that Perplexity would “definitely” sell such a device if the post got more than 5,000 likes.
StabilityAI adds new capabilities to Stable Diffusion 3.5 Large with the release of three ControlNets: Blur, Canny, and Depth. Blur enables achieving extremely high fidelity upscaling, including resolutions of 8K and 16K and is perfect for tiling low-resolution images into large, detailed visuals. Canny edge maps allow users to structure their generated images. Depth maps, generated by DepthFM, enables users to guide image generation.