AI news

The latest from the world of artificial intelligence. Daily monitored sources, research, product and company news.

Sections

Latest articles (50 of 424)

OpenAI temporarily lifts the five hour usage limit in ChatGPT Work and Codex — OpenAI has temporarily removed the five hour usage limit in ChatGPT Work and Codex for Plus, Business and Pro plans. It also reset everyone's usage counter and promises that GPT-5.6 Sol will consume less for the same work. It sounds like a clean win, but the key word in the whole announcement is "temporarily". And that is exactly what matters most for companies building processes on these tools.
Claude Science: Anthropic Introduces a Dedicated AI Workbench for Scientists and Researchers — Anthropic has introduced Claude Science, a specialized workspace designed specifically for scientific research. The new platform consolidates fragmented databases, analytical tools, and computational resource management into a single application. The goal is to streamline complex processes, from data analysis to preparing materials for publication. How does this new tool, currently in beta, work in practice, and who can access it?
The bank account just got AI, and it takes money management further — A bank account usually just passively holds money, and you have to click through menus, export data and switch between tools. The American fintech Mercury is flipping that. Its new feature, Command, is artificial intelligence built directly into the account. You tell it in plain language what you need and it does it, but only with your confirmation. For the Czech market it is an interesting look at where money management is heading.
Swap the question of AI's return for the question of what waiting costs you — For decades financial decisions rested on certainty. Proof first, then action. With the arrival of AI that stops holding, according to IBM. Waiting until the numbers are clear is now more of a risk, because competitors who did not hesitate gain a head start in the meantime. So it is time to swap the question of what the return is for the question of what waiting costs us. I sum up what follows from it for smaller Czech companies too.
Ford bet that AI would replace experienced people, then spent three years fixing the mistake — Ford climbed to the top of the JD Power quality ranking, but the road there ran through an uncomfortable lesson. The automaker bet on artificial intelligence and automation in production and underestimated its most experienced engineers. Some of them left before the models could be trained. Ford then had to hire hundreds of experienced people back over three years to fix the mistakes and to actually teach the AI.
The era of spending on AI at any cost is ending, and customers are leaving OpenAI and Anthropic — For two years the rule was simple. Run everything through the most powerful model and do not worry about the price. Now that is breaking. According to CNBC, companies are starting to rein in their AI bills, looking for efficiency, and some customers are leaving OpenAI and Anthropic for cheaper models. For both companies, which are heading for the stock market, it may be the end of their fastest growth. And for your company, a clear lesson in how to approach AI.
Anyone can send emails today, but do you know Resend? — Sending emails from an app is now a matter of a few lines of code, and the Resend service handles it elegantly. But sending an email is one thing and doing it right is another. A new report found that of more than 376,000 emails, only a tiny fraction passes a basic accessibility test.
First Anthropic, now OpenAI. Washington is blocking the next most powerful AI — The most powerful language models are starting to behave like sensitive technology under state supervision. According to the Financial Times, OpenAI is to release its new GPT 5.6 model in stages, with the US government approving its first partners. Just days earlier, export controls hit Anthropic and its Fable and Mythos models. For companies, the signal is clear about how fragile access to the best AI has become.
Czech President vs. government. What does AI think about it? — Beer consumption in Czechia is falling to a historic low, and with it the pub debates where political opinions were always tested. So part of the public is starting to ask artificial intelligence. I tried it. I put the same question about the dispute between the president and the government over the trip to Ankara to five tools. But the result kept changing depending on how I asked and which model I chose.
The most powerful AI models are entering the same regime as military technology — The US government ordered the most advanced publicly available AI model pulled from the market just three days after launch. The reason was national security. For companies that build their processes on rented intelligence, it is a signal that access to the best models is no longer a given. What does it mean for company strategy, and where should they look for alternatives?
File Generation in Gemini, Smarter Sheets, and More Google Workspace Updates — Last month, Google added several new features to Workspace. Some are already available in Czech, while others are still waiting for localization. Generating files directly from Gemini chat is a welcome step, though Claude has been capable of LaTeX output since autumn 2024. I reviewed all the updates and included a practical use case for each.
Dreaming Feature Gives AI Agents Space to Reflect Between Tasks — At the Code with Claude conference, Anthropic introduced a new capability for Claude agents called Dreaming. It essentially allows the AI to review its own sessions, identify recurring errors, and extract actionable lessons from them. For companies deploying agents on longer work tasks, this could be a significant step toward more reliable deployment.
In China, You Can No Longer Fire an Employee Over AI — Two Chinese courts in Hangzhou and Beijing ruled that a company cannot dismiss an employee simply because AI has taken over their work. According to the courts, deploying AI is a strategic decision, not an unforeseen change in circumstances. Meanwhile, the global tech industry laid off over 78,000 people in just the first four months of 2026 — nearly half of them because of AI. The US and EU provide no comparable protection.
Hedy AI Can Handle Your Entire Meeting Analysis Directly on Your Device — Most AI meeting assistants send your conversations to external servers. Hedy AI takes the opposite approach. It can now handle the entire analysis, summaries, and live notes directly on your computer or iPhone. And it is currently available for a one-time payment on AppSumo instead of a recurring subscription.
Czech E-shops Are Mass-Testing AI, But Only a Fraction Manages It Strategically — Almost all Czech e-shops are already experimenting with artificial intelligence. However, a new study reveals that only 6 percent of them have managed to truly integrate AI into their processes and management. How are Alza.cz, PPL, and Zásilkovna working with it, and what is preventing others from moving from experiments to strategy?
Adobe Is Deploying AI Across All Its Applications. What Can Photoshop, Illustrator, and Premiere Do Today? — Adobe is gradually integrating artificial intelligence throughout the Creative Cloud ecosystem. From a conversational assistant in Photoshop to generative vector graphics in Illustrator and automatic masking in Premiere. What exactly can each application do today, and how is it changing the work of creatives?
Companies Cut Jobs Over AI, Then Turn Around and Hire New People — Tech companies are laying off workers en masse and pointing to artificial intelligence. On closer inspection, however, work is not disappearing — it is changing form. While big players are shifting resources to contractors and flexible arrangements, Czech companies like Etnetera show how AI agents are transforming internal processes. And Direct Group is planning to cut a third of its workforce. What does this mean for employees and employers?
How to safely use Claude in Chrome without losing control of your browser — The Claude in Chrome extension lets an AI assistant work directly in your browser. It clicks, fills out forms, and switches tabs. But that also comes with risks — from unwanted actions to prompt injection attacks. Here is a practical guide on how to set up and use the extension so you stay in control.
Five AI agents that want to control your computer. The big comparison of 2026 — 2026 brought a new category of artificial intelligence. AI agents no longer just answer questions — they directly control your computer, organize files, write code, and complete tasks even when you're away from the keyboard. This article compares five major players: OpenClaw, Meta Manus, Perplexity Computer, Claude Cowork with Dispatch, and Microsoft Copilot. Who offers what and at what price?
Gemini is settling into Google Workspace and changing how you work with documents, spreadsheets, and presentations — In March 2026, Google launched a beta of a major Gemini integration into Docs, Sheets, Slides, and Drive. Just describe what you need and Gemini will reach into your emails, files, and chat to prepare a first draft of a document, spreadsheet, or slide. How does it work in practice and what does it mean for everyday work?
Voice Deepfakes Don't Fool Your Auditory System — Only Your Conscious Mind — New research shows that people cannot consciously distinguish an AI voice from a human one. But the brain manages it on its own. Using EEG, neuroscientists discovered that after a brief training period, the auditory system begins to "tag" fake voices differently from real ones — long before a person consciously registers it. What does this mean for the future defense against voice fraud?
Czech lawyers and artificial intelligence. A fresh survey brings surprising numbers — More than 60% of Czech lawyers use artificial intelligence tools daily. Half would pay a maximum of 1,500 CZK per month for a specialized AI solution. A fresh survey of 123 legal professionals reveals exactly what they do with AI, what they fear, and what they expect from it in the future.
Artificial intelligence is becoming the invisible ally of an aging population — One in four people over 65 falls every year. Loneliness increases the risk of dementia comparably to smoking fifteen cigarettes a day. And yet technologies already exist today that can detect a fall without cameras, remind people to take their medication in a human voice, or reduce feelings of loneliness by 95%. Artificial intelligence is moving from an abstract concept to a concrete helper for millions of seniors around the world.
Clarivate AI50 reveals who is really driving AI innovation in the world — Analytics firm Clarivate has published its new AI50 ranking, which uses patent data to identify 52 organizations with the strongest AI inventions in the world. Unlike conventional rankings, it is not concerned with revenue or company size, but with measurable quality and global innovation intent. The result? Four distinct models of AI leadership and a surprising finding about where real value concentrates.
Copilot is for entertainment purposes only, says Microsoft — Microsoft has spent years investing billions in its AI assistant Copilot and integrating it into Windows and Office applications. Yet in its terms of use, it states in black and white that Copilot is for entertainment purposes only and that users should not rely on it for important decisions. How should we make sense of this contradiction?
Google's Gemma 4 runs on your phone and beats models twenty times its size — Google DeepMind has unveiled Gemma 4, its most capable family of open AI models to date. Four variants cover everything from smartphones to powerful servers, work with text, images, video, and audio, and are available under the fully open Apache 2.0 license. The largest model ranks among the top three open models in the world on benchmarks.
Why Anthropic cut off OpenClaw and other third-party tools from Claude subscriptions — From April 4, 2026, Anthropic no longer allows Claude Pro and Max subscribers to use third-party tools within their flat-rate plan. The decision affects over 135,000 developers and for some means a cost increase of up to fifty times. What is behind this, how does it affect regular users, and what does the credits offer email that many subscribers recently received actually mean?
Seven AI breakthroughs in March 2026 that are pushing the boundaries of artificial intelligence — March 2026 brought a fundamental shift in artificial intelligence. AI systems are no longer just text generators — they are transforming into autonomous agents that plan, decide, and act. The cost of running models is falling, robots are learning to operate in the real world, and language models are beginning to understand code and security at the level of experienced developers. Here is an overview of the seven most important breakthroughs of the month.
The Ignite AI Ambitions conference showed that the Czech Republic has talent and energy, but needs the courage to scale — The third edition of the Ignite AI Ambitions conference drew over 800 guests from business, government, academia, and the technology sector to Prague's Divadlo Hybernia. The flagship event of the Czech Artificial Intelligence Association raised the question of what role the Czech Republic wants to play in the global AI race — and whether it has the courage to truly fulfill that ambition. We were there.
Czech AI detector beats foreign competition — Brno-based startup DayOne Labs has unveiled a tool called Veritas.cs that allegedly recognizes with 99.7% accuracy whether a text was written by a human or artificial intelligence. If the results are confirmed, it would be the most accurate detector in the world.
The Lazy Millionaire's Manual: 10 Ways to Get Rich on the Road with AI and Cleverness — You know the cliché that there's no such thing as a free lunch? In the age of artificial intelligence and global connectivity, that's not entirely true anymore. Imagine a reality where your bank account grows while you sip coconut on a Thai beach or explore the hidden alleys of Rome. No fraudulent scheme, no risky stock market bets. The key is the "smart lazy" approach — deep research, modern technology, and opportunities others miss because they're too busy. If you have time to think and a taste for letting algorithms — or chance — work for you, you're halfway there. Here are 10 strategies for trading hard grind for smart strategy.
Claude Dispatch and Computer Use let you control your AI assistant from your phone while it works on your computer — Anthropic has pushed its AI assistant Claude a significant step further. The new Computer Use feature allows it to click, type, and navigate applications directly on your computer. Combined with the Dispatch feature that connects the desktop with a mobile phone, this creates an AI agent you assign a task to from your phone — and it completes the task on your computer even when you're not sitting at it.
Spain is showing Europe how quickly you can catch up with leaders in AI adoption — While Germany and the UK lead in overall maturity, Spain is the fastest-growing AI market in Western Europe. A new report from Redegal reveals what is driving this jump, why 2026 represents a tipping point, and what impact artificial intelligence is having on e-commerce, marketing, and customer service.
Microsoft's MAI-Image-2 is heading to offices and Czech companies should take notice — On March 19, 2026, Microsoft unveiled MAI-Image-2, its own image generation model. It is now part of both Office and Teams, handles Czech diacritics, and has outperformed Midjourney v6.1 in global comparisons. For Czech companies, this means cheaper graphics, better data security, and an end to dependency on external tools.
Nearly a quarter of job listings in the EU already require AI skills — One in four entry-level job listings in the EU today requires knowledge of artificial intelligence. Half of small and medium-sized businesses say their employees lack AI skills. Google is responding with a new initiative called AI Works for Europe and a $30 million fund to help students, employees, and companies across Europe, including the Czech Republic.
The TRUMP AMERICA AI Act aims to unify rules for artificial intelligence across the United States — Senator Marsha Blackburn has introduced a nearly 300-page bill that would create a unified federal framework for regulating artificial intelligence in the US for the first time in history. The bill addresses developer liability, child protection, copyright, and the elimination of existing online platform protections. Here is an overview of what the proposal contains and what it could mean for the rest of the world.
Data Is the Weak Link in Enterprise AI and Startup Validio Is Building a Multi-Million Business on It — Most corporate AI projects never reach production. Not because of poor models, but because of low-quality data. Stockholm-based startup Validio bet on this back in 2019 and has now raised $30 million to help companies prepare their data for the AI era. Why is data quality so crucial and what exactly does Validio offer?
Musicians can finally fight back against AI clones of their voices — Artificial intelligence can now clone a singer's voice from just a few seconds of recording. Researchers at Binghamton University have developed a tool called My Music My Choice that adds inaudible changes to songs. A listener hears no difference, but cloning software produces only distorted noise. Protection arrives before the fakes do.
How to switch from ChatGPT to Claude and why millions of people are doing it right now — Anthropic has launched a tool to import memory from competing chatbots into Claude. It arrived at the moment when millions of users are leaving ChatGPT over OpenAI's controversial deal with the Pentagon. As a result, Claude has climbed to the top of the App Store and new sign-ups have quadrupled. What is behind it and how does the switch work?
Google's Nano Banana 2 Combines Local AI with Cloud Power and Targets Creators and Small Businesses — Google has presented the second generation of its compact AI device Nano Banana 2. Built on the Gemini Flash model, it promises professional generative AI features completely free of charge and aims to directly compete with paid alternatives. How does it all work, what can it do, and where are the open questions?
Microsoft's Copilot Tasks Finally Does the Work Instead of Just Talking About It — Until now, chatbots have mainly given us advice and generated text that we still had to finish ourselves. Microsoft is now trying something different. Copilot Tasks is an AI assistant that opens its own browser, browses the web, unsubscribes you from newsletters, drafts email replies, or monitors apartment listings. You just tell it what you need. It does the work. How does it work and what does it mean for the future of working with AI?
Why Local AI Alone Doesn't Guarantee Your Security — Many companies think that running AI locally is enough to take care of data security. But the reality is more complex. Local operation does mean that data doesn't leave your network, but without proper design, clear rules, and ongoing oversight, even a local model can become a security risk. Let's look at what local AI can actually do, where its limits are, and how to approach it with a clear head.
Why Choosing the Right AI Model Now Matters More Than Knowing How to Prompt — The era of needing to be an expert at prompting is slowly coming to an end. The best AI models today guide you to the result themselves. But beware — choosing the right model is more important than ever. A simple car wash test showed that even a trivial logical task is reliably solved by only 5 out of 53 tested models. Which one should you choose and why does it matter?
LLM, RAG, AI Agent, and Agentic AI — Four Layers of Artificial Intelligence Worth Understanding — Language model, RAG, AI agent, agentic system — it sounds like four different things, but in reality they are layers of one ecosystem. In this article, we'll explain what each layer does, where one ends and another begins, and most importantly, when you need each one. Clearly and without clichés.
9 Proven Frameworks That Will Help You Write Better Prompts for Artificial Intelligence — Most people write prompts blindly — they write a sentence, hope for the best, and wonder when the AI responds off-target. Yet it only takes a simple framework to give your prompt a clear structure. In this article, we'll go through 9 specific prompt frameworks with practical examples, and at the end we'll add one bonus that will help you think about AI from a completely different angle.
From a One-Hour Prototype Came 1.5 Million AI Agents and an Offer from OpenAI — All it took was one hour of work to create a prototype that became a platform with 1.5 million AI agents. Austrian developer Peter Steinberger found himself courted by the heads of Microsoft, Meta, and OpenAI. Why did Sam Altman win, what can AI agents actually do, and where are their biggest weaknesses so far?
At Accenture, Results Are No Longer Enough — Managers Who Don't Work with AI Won't Get to Leadership Positions — Consulting giant Accenture has added a new condition to its promotion criteria — regular use of internal AI tools. The company has started tracking individual logins to AI systems for senior managers and is including this data in this year's career advancement decisions. This is one of the first cases where a major corporation explicitly links career growth to artificial intelligence adoption.
WordPress.com Has a New AI Collaborator That Understands Your Website — WordPress.com is introducing an AI assistant that works directly in the editor, media library, and team chat. Unlike external tools, it understands the structure and content of your website, making changes on the spot and in real time. No copying, no prompting outside the platform. Just describe what you want and the assistant will do it for you.
China Is Deploying Artificial Intelligence Against Corruption in Public Procurement — China has published guidelines under which artificial intelligence is to monitor public procurement and tenders. The goal is to uncover backroom deals, bid manipulation, and corruption — something that human officials simply cannot keep up with given the enormous volume of contracts. AI in China already has its first real catch: based on data analysis of one city project, it detected suspicious connections that led to the arrest of a state administrator. What exactly should AI systems be able to do, and how far can digital surveillance go?
The More You Let AI Think for You, the More You Forget to Think for Yourself — Artificial intelligence can help you write an email, prepare an analysis, and propose a strategy. But what happens when your AI tool isn't available tomorrow? Research shows that long-term reliance on AI leads to so-called deskilling — a silent loss of skills that most users don't even notice. And it applies to you too.