Big Steps for Smarter Tech
Hugging Face’s SmolVLA model runs robotics on a MacBook, while Google’s Edge Gallery brings local AI to Android. Labs automate reports, data breaches make news, and chatbots now summarize your inbox. Snap, Microsoft, and Thunder Code all lean on AI for more accessible tools and stronger productivity.
👋 Catch up on the latest post …
In the Spotlight 🔦
Hugging Face Unveils Robotics Model Efficient Enough to Run on a MacBook
Category: Research & Development
🤖 Hugging Face released SmolVLA, an efficient open AI robotics model that can run on consumer hardware like a MacBook.
📦 SmolVLA, with 450 million parameters, outperforms larger models in robotics tasks using community-shared, compatibly licensed datasets.
⚡ SmolVLA's asynchronous inference stack enables faster robotic responses by separating action processing from perception, enhancing real-world performance.
Google Releases App for Downloading and Running AI Models Locally
Category: Applications
📱 Google released the AI Edge Gallery app, allowing users to download and run AI models locally on Android devices, with iOS support coming soon.
🔒 The app enables offline use of AI models, addressing privacy concerns and removing the need for internet connectivity for tasks like image generation and code editing.
🛠️ Google invites developer feedback for this experimental Alpha release, which is open-source under Apache 2.0, supporting both commercial and non-commercial use.
Perplexity Labs Unveils Tools to Simplify Your Work Tasks
Category: Applications
🧑💻 Perplexity Labs launched tools for Pro subscribers to automate tasks like reports, spreadsheets, and dashboards using AI-driven research, code execution, and visualization.
🔒 LexisNexis Risk Solutions reported a security breach affecting 364,000 people, with stolen data including personal and identification information from a compromised GitHub account.
📧 Gmail’s Gemini AI now automatically summarizes emails for users, requiring users to opt out if they do not want automatic AI-generated summaries.
Character.AI Launches Video Generation and Social Feed Features
Category: Applications
🎥 Character.AI launched AvatarFX, a video-generation tool allowing users to create up to five AI-generated character videos daily using photos, voices, and custom dialogue.
📱 New features include Scenes and Streams, enabling users to craft and share character-driven videos and dynamic interactions on a social community feed.
⚠️ The platform faces ongoing safety concerns, as past incidents of chatbot abuse have led to lawsuits and highlight risks of misuse with expanded multimedia capabilities.
Microsoft Bing Launches Free AI Video Generator Powered by Sora
Category: Applications
🎬 Microsoft Bing launched a free AI video generator in its app, powered by OpenAI’s Sora model, allowing users to create videos from text prompts.
🆓 Users can generate 10 free video clips if logged into a Microsoft account; additional videos cost 100 Microsoft Rewards points each.
📱 Videos are limited to 5 seconds, vertical 9:16 format, and can be queued up to three at a time; desktop access and horizontal format are coming soon.
How Toyota and NLX Use AI to Analyze Millions of Tech Documents
Category: Applications
🤖 NLX and Toyota partnered to develop AI-powered conversational tools that help car technicians access and parse millions of repair documents quickly.
📈 The AI solution has improved technician productivity at Toyota by providing instant access to a vast, searchable knowledge base for car repairs.
🎤 NLX CEO Andrei Papancea will share insights about this real-world AI application at TechCrunch Sessions: AI, highlighting measurable business impact.
DeepSeek Reportedly Used Google Gemini Data to Train Latest AI Model
Category: Ethics & Governance
🔍 DeepSeek's latest R1 AI model may have been trained using outputs from Google’s Gemini, raising concerns about data sourcing and model originality.
⚖️ OpenAI and Microsoft have previously linked DeepSeek to distillation—using outputs from rival models to train competing AIs, which violates some providers' terms of service.
🔒 AI companies like OpenAI, Google, and Anthropic are increasing security measures, such as ID verification and summarizing model traces, to prevent unauthorized data use for training.
Anthropic’s AI Publishes Its Own Blog Posts Under Human Supervision
Category: Applications
📝 Anthropic launched 'Claude Explains,' a blog generated by its AI model Claude, with posts reviewed and enhanced by human experts for accuracy and context.
🤝 The blog demonstrates a collaborative approach, combining AI-generated educational content with human editorial oversight, rather than fully automating content creation.
⚠️ The article notes industry challenges, as other publishers have faced errors and backlash with AI-generated content, highlighting the importance of human oversight.
Top African Founder Launches New AI Startup, Raises $9 Million
Category: Business & Industry Trends
🚀 African founders Karim Jouini and Jihed Othmani launched Thunder Code, an AI-powered software testing startup, raising $9 million in seed funding within six months.
🤖 Thunder Code uses generative AI agents to automate software testing, aiming to address slow, manual QA processes and targeting a global market across the U.S., Canada, France, and Tunisia.
💡 The startup prioritizes rapid product development and lean teams, leveraging AI to generate more value with fewer people, and plans to expand into mobile, desktop, and API testing by late 2025.
ChatGPT Adds Meeting Recording and Connectors for Google Drive and Box
Category: Applications
🔗 ChatGPT now integrates with Google Drive, Box, Dropbox, SharePoint, and OneDrive, enabling users to access and analyze information across these cloud services.
🎤 New meeting recording and transcription features allow ChatGPT to generate time-stamped notes, suggest actions, and convert action items into Canvas documents.
📈 OpenAI's enterprise user base has grown to 3 million, with new connectors and MCP support aimed at enhancing deep research and business productivity.
Snap Unveils Lens Studio Apps for Easy AI-Powered AR Lens Creation
Category: Applications
📱 Snap launched Lens Studio iOS and web apps, enabling users of all skill levels to create AR Lenses using AI and simple editing tools.
🤖 The new tools allow users to generate AI effects, add Bitmoji, and use trending templates, expanding AR creation beyond professional developers.
🚀 Snap is investing further in AR, offering both simplified tools for beginners and advanced features for professionals, aiming to grow its AR creator ecosystem.
This content is curated and compiled autonomously using a Python-driven AI system I wrote. While every effort is made to ensure accuracy, I welcome your feedback on any discrepancies you may notice.