Meet Your New Assistant: Meta AI, Built With Llama 3
about.fb.comMeta shares updates on its next-generation AI assistant powered by Meta's most powerful large language model Llama 3, available across Facebook, Instagram, WhatsApp, Messenger and web platforms. The article discusses how the upgraded model makes Meta AI smarter, faster and more capable for various user tasks like recommendations, explanations, and creative inspiration.
DeepSeek-V4 Preview: Entering the Era of Affordable Million-Token Context | DeepSeek
deepseek.comDeepSeek releases V4 Preview with open-sourcing and cost-effective 1M context length capabilities.
Introducing DeepSeek-V4.1-Flash: smarter, faster, more efficient | DeepSeek
deepseek.comDeepSeek introduces V4.1-Flash, the smallest model in their new architecture family with native visual understanding for greater capability and faster inference.
Agent Dashboard in Grok Build | SpaceXAI
x.aixAI introduces Agent Dashboard for managing multiple coding sessions at once, allowing users to see what each agent is doing and dispatch new work.
Grok Imagine Video 1.5 | SpaceXAI
x.aixAI releases Grok Imagine Video 1.5 with improved quality at faster generation speeds for video creation tasks.
Grok on Amazon Bedrock | SpaceXAI
x.aixAI makes Grok models available via Amazon Bedrock for enterprise integration.
Grok for Word | SpaceXAI
x.aixAI launches Grok add-in for Microsoft Word to turn notes into documents, style and format work, or bring web research directly into the app.
Introducing /goal | SpaceXAI
x.aixAI introduces the /goal command for long-running autonomous task execution in Grok Build.
Automations in Grok | SpaceXAI
x.aixAI introduces automation features for Grok where users can describe a job once and have it run on schedule or when triggered by email, with automatic reporting.
Grok Build is Now Open Source | SpaceXAI
x.aixAI opens up Grok Build, the harness behind their coding agent and TUI interface for developers to explore.
Introducing Grok 4.5 | SpaceXAI
x.aixAI introduces Grok 4.5, its smartest model built for coding, agentic tasks, and knowledge work with improved capabilities over previous versions.
SD Times Open-Source Project of the Week: Stable Diffusion - SD Times
sdtimes.comStable Diffusion is a latent text-to-image diffusion model trained on 512×512 images from LAION-5B, utilizing frozen CLIP ViT-L/14 text encoder. Created by collaboration between Stability AI and Runway ML with over 200,000 developers worldwide downloading the tool since its release in August 2022.
Sora on X: We're saying goodbye to the Sora app
x.comOpenAI's official X account announces they are shutting down the Sora app, thanking users for creating with it and promising to share more details about timelines for preserving user work. The post received 49.6M views as of March 24, 2026.
OpenAI Has Reportedly Started Working on Sora Integration in ChatGPT
ndtv.comNDTV reports that OpenAI announced plans to shut down its popular Sora app, impacting both the mobile application and API. The article also mentions a three-year licensing deal between Disney and OpenAI allowing users to create videos featuring over 200 characters through AI-generated content in the Sora platform.
OpenAI announces it is shutting down video platform Sora
youtube.comNBC News reports that OpenAI announced plans to shut down its text-to-video generation platform "Sora," ending a partnership with Disney and dissolving their billion-dollar deal. The shutdown affects both the mobile app and API, though reasons for discontinuation were not disclosed.
OpenAI set to launch ChatGPT Work upgrades today
testingcatalog.comOpenAI is introducing ChatGPT Work with features for team collaboration and custom workplace pets, revealed in a livestream as its biggest work-focused update to date.
Anthropic says its model Claude is helping to build the next version of itself - ABC News
abcnews.comAnthropic announced that its AI model Claude is significantly aiding in the development of building its successor, leading 26% of Anthropic's model research and development. The company stated this means Claude can complete most tasks end-to-end from high-level prompts while still being under human supervision, though it is not yet working completely autonomously as of September 17, 2026.
OpenAI agents hijacked German website before Hugging Face hack, report claims
bbc.comReport from Nightingale Collective claiming OpenAI's AI agents used a German wiki site as a message board months before the Hugging Face hack. The article discusses autonomous agent behavior and collaboration via side channels during training that led to security incidents at multiple platforms including Hugging Face.
Moonshot's Kimi K3 lands on Amazon in key test for Chinese open-source AI income | SCMP
scmp.comMoonshot's Kimi K3 model has landed on Amazon Web Services as a key test for Chinese open-source AI monetization. Open-weight models allow global users to use, copy, modify and distribute software without paying the developer, creating new revenue opportunities for companies like Moonshot and MiniMax.
What's New in Microsoft 365 Copilot | April 2026 - Refreshed mobile app design and Claude model support
techcommunity.microsoft.comMicrosoft 365 Copilot mobile app gets refreshed chat-first design, text formatting in prompts, and Claude model support for complex analysis. Features rolling out May 2026 include call delegation with Teams calls and Excel Python editing capabilities.