AI

A global QNX study of 1,000 robotics developers reveals a paradox: 95% demand real-time behavior, yet 91% rely on general-purpose operating systems. Certification delays 66% of projects, and only 29% of developers feel confident making safety-critical decisions in real-world environments.

Alibaba Cloud launched Qwen Cloud in Singapore on May 26 — the first cloud platform built from the ground up for AI agents. It offers three entry points (Skills for agents, CLI for workflows, Web for humans), six modalities including video and speech, and starts for free with no credit card required.

Nous Research has unveiled NousCoder-14B. This specialized coding model can go head-to-head with even the best proprietary systems like Claude Code, while remaining fully open for the community.

While banks worldwide are rapidly deploying autonomous AI systems to approve transactions or detect fraud, traditional oversight mechanisms are failing to keep up. According to an HSBC expert, a single hallucination by an agentic model could trigger an unauthorized payment before anyone can stop it.

Agentic artificial intelligence could speed up cybersecurity operations by up to 365x, according to a new study by MSSP Security Consulting. But the reality is sobering: 95% of security operations centers worldwide lack the infrastructure, processes, or trust to deploy autonomous AI agents.

Agentic artificial intelligence could unlock up to $550 billion in value across the global real estate sector, according to McKinsey. The key isn't deploying isolated tools, but completely rebuilding company processes.

Generative artificial intelligence has reached 53% population adoption in just three years — faster than personal computers or the internet. The gap between American and Chinese models has shrunk to just 2.7%, while data center energy consumption is surging and security incidents are on the rise.

Xiaomi is slashing prices for its MiMo-V2.5 series AI models by as much as 99 percent. Unified billing regardless of context length, higher token limits, and free speech synthesis — all starting May 26. China's AI price war is escalating.

Microsoft introduces "computer use" capability for Copilot Studio. Agents can now control websites and desktop applications just like humans, opening the door to automation even in systems without modern interfaces.

Anthropic, the creator of Claude AI, grew into a company valued at $380 billion in just five years. How does constitutional AI work, what can the latest Mythos model do, and why did the company dare to say no to the Pentagon? A complete guide for Czech readers.