AI News

The father of deep learning, Yann LeCun, has left Meta and, with his Parisian startup AMI Labs, raised a staggering $1.03 billion. He's betting it all on the idea that large language models are a dead end.

Meta just dropped Llama 4 Scout and Maverick — open-weight models boasting a record-shattering 10 million token context and native multimodality. The catch? The EU is explicitly excluded from the license, completely blocking Czech developers from deploying them on their own infrastructure.

Dive into the crucial themes emerging from global leaders' discussions. From the rise of physical AI and the OpenClaw standard to the profound economic impacts on Software as a Service and escalating geopolitical tensions.

Genspark just unveiled its groundbreaking Desktop AI Agent, designed to directly interact with your computer applications. Get ready to see how this new tool could revolutionize the way you work on both Windows and macOS!

Anthropic has made a controversial decision: their latest AI model, Claude Mythos, won't be released to the public. We're diving deep into the reasons behind this shocking move and what it means for AI safety and developers in Czechia.

SambaNova and Intel have unveiled a plan for heterogeneous inference — a trio of specialized chips designed for agentic AI. The new SN50 RDU promises to be 5x faster than Nvidia's B200, all while offering 8x lower costs.

We're moving beyond AI that just answers questions, into an era where it autonomously tackles complex tasks. Discover what the rise of "agentic" artificial intelligence means for businesses and everyday users.

Boston's IT leaders are unveiling their open-source agentic AI tools! The goal? To boost government transparency and empower other cities to leverage proven technologies without relying on big corporations.

Anthropic just redefined AI agents with its new Cowork feature for Claude Desktop. This game-changing tool allows the model to directly work with your local files – no coding required – representing a massive leap in AI interaction.

Anthropic has just unveiled Claude Mythos, the most powerful cybersecurity AI model ever created — but they're deliberately keeping it from the public! Only 40 hand-picked companies, including tech giants like Apple, Google, and Microsoft, have access to it through Project Glasswing.