Category AI News

Twin partners with Qonto to automate invoice retrieval using AI. Their new agent speeds up the process, supporting thousands of apps in just months. Paris-based AI startup Twin has launched its first automation agent in partnership with fintech company Qonto. The tool, called Invoice Operator, helps businesses automatically retrieve invoices, saving time for Qonto’s 500,000+ customers. Invoice Operator identifies missing invoices, accesses relevant services, and downloads them. If login credentials are needed, the agent pauses for manual input before continuing its task. Once retrieved, invoices are attached to transactions in…

FOSS developers combat AI web crawlers with inventive tools like Anubis and Nepenthes. These solutions help block bots that ignore robot.txt files. AI web crawlers are causing havoc for developers, especially in the open-source community. Many crawlers ignore the robot.txt file, which tells bots what to avoid. This leads to issues like server crashes and slowdowns. Open-source developers are fighting back with creative solutions.FOSS developers combat AI web crawlers with inventive tools like Anubis and Nepenthes. These solutions help block bots that ignore robot.txt files. Xe Iaso, a developer, created…

OpenAI pulled GPT-4o’s image feature from free ChatGPT users after a surge in Studio Ghibli-style AI art. Paid users still have access. No return date announced. OpenAI has removed its upgraded AI image generator from the free tier of ChatGPT. The tool, powered by GPT-4o, was launched to improve realism and text clarity in AI-generated images. Why OpenAI Pulled the Feature The decision came just a day after release. Users flooded social media with AI art mimicking Studio Ghibli’s style, recreating famous scenes and memes. Even CEO Sam Altman joined…

Learn how to use Apple’s Image Playground AI art tool on iPhone, iPad, and Mac to create and edit stunning images from text prompts and photos. Apple has launched Image Playground, an AI-powered art generator for iPhones, iPads, and Macs. It turns text prompts into images, making it easy to create unique visuals. How to Use Image Playground If you have iOS 18.2, iPadOS 18.2, or macOS Sequoia 15.2, Image Playground is preinstalled. It requires Apple Intelligence, which should be enabled by default. Refining Your AI Art Swipe through results…

Hugging Face calls on the US to support open-source AI, highlighting its role in innovation, security, and economic growth. Read key policy recommendations. Hugging Face has urged the US government to prioritize open-source AI in its upcoming AI Action Plan. The company argues that open-source models drive innovation, economic growth, and security. Hugging Face’s platform hosts over 1.5 million public AI models, serving seven million users. It proposes three key pillars for AI development: strengthening open-source ecosystems, ensuring efficient adoption, and promoting security standards. The company stresses that open-source AI…

Explore OpenAI’s GPT-4o voice models, enhancing transcription accuracy and realistic text-to-speech for building reliable AI agents and chatbots. OpenAI has introduced new AI models for transcribing speech and generating voice in its API. These models improve significantly over previous versions. The goal is to help developers easily build AI “agents,” automated systems that perform tasks for users, like customer service chatbots. The new text-to-speech model, called “gpt-4o-mini-tts,” creates speech that is more realistic and flexible. Developers can guide the model using natural language instructions. For example, they can request the…

Anthropic’s Claude 3.7 Sonnet is the latest iteration of the Claude AI assistant, bringing a significant upgrade in how the model reasons and interacts. Unveiled in late February 2025 (https://www.anthropic.com/news/claude-3-7-sonnet), Claude 3.7 (code-named “Sonnet”) is described as Anthropic’s most intelligent model to date​. What sets Sonnet apart is its unique ability to operate in two modes: a standard mode for fast, conversational answers and a new extended thinking mode for step-by-step, in-depth reasoning. In this article, we’ll summarize the latest enhancements of Claude 3.7 Sonnet, its applications (from coding to…

Microsoft’s Muse AI is a pioneering generative model designed not for text or images, but for video games. Unveiled in early 2025 (https://www.microsoft.com/en-us/research/blog/introducing-muse-our-first-generative-ai-model-designed-for-gameplay-ideation/), Muse can learn how a game works and then generate new gameplay – both the visuals and the player actions – on its own​. In essence, Muse is an AI that can watch a game, understand its mechanics and world, and then imagine “what happens next” in that game. This represents a novel use of AI: rather than creating a static image or text, Muse creates interactive…

Google DeepMind’s Veo 2 is the latest leap forward in AI-generated video, offering ultra-high-resolution output and improved realism. Announced in late 2024 and gaining attention through early 2025, Veo 2 is a text-to-video model that can create video clips in a wide range of styles – now up to 4K resolution and extended lengths of several minutes​. This next-generation tool is Google’s answer to models like OpenAI’s Sora, and it aims to one-up the competition with more sophisticated understanding of scenes and cinematography. Let’s explore the new developments and features…

OpenAI’s Sora Europe Access rollout marks the long-awaited arrival of its text-to-video tool in European countries. Sora is a cutting-edge AI model that transforms text prompts (and even images) into short video clips. It launched in the US in late 2024, and now – about three months later – OpenAI has expanded access to users across the EU and UK​. This expansion opens the door for European creators to experiment with AI-generated videos, albeit with some conditions and limitations. Contents Sora Finally Reaches European Users After an initial US launch,…