The basis for this series is the biweekly LinkedIn article by Tom Van ‘t veld, Learning Innovator at OASE (powered by Xylos). Tom closely follows AI developments and has been sharing his findings for some time. On the Xylos blog, we bring you a selection of the most relevant insights, focused on what matters to organizations and the people working in them. Welcome to edition one.
EDITION 1 – APRIL 13, 2026
No player dominated the news as much as Anthropic over the past two weeks, and not always in a positive sense. But Google, Microsoft and the European regulators didn’t let their guard down either. Here are the six stories that really stood out.
Anthropic had an eventful week
Claude Cowork left the preview and is now generally available to enterprises. Tom’s tests were impressive, especially considering that this is just the end-user layer. OpenAI quickly countered with a new $103-a-month ChatGPT Pro subscription specifically aimed at programmers. The battle for the professional market is in full swing.
At the same time, Anthropic announced Claude Mythos, and that announcement is … extraordinary. The company had an outside psychiatrist conduct 20-hour psychodynamic sessions with the model, because Anthropic does not rule out the possibility that very powerful AIs have something resembling an inner experience. Whether that was really necessary or rather a successful way to make headlines is hard to say. The rest of the message does sound familiar: as with OpenAI’s GPT-2 in 2019, the model will not be released publicly for now because it would be “too dangerous. We’ve seen this movie script before.
Then there was the source code. Anthropic accidentally made the full source code and architecture of Claude Code publicly, suddenly giving competitors a detailed look into one of their core products. Through thousands of copyright takedowns, the company tried to have the code removed from GitHub, but developers rewrote it in other programming languages and republished it. A piquant detail also surfaced from the leak: Claude Code explicitly records signs of user frustration, such as swear words and negative comments, while erasing references to itself from the generated code so that it appears more human.
Mythos as social risk: dramatic or justified?
Claude Mythos’ announcement also caused a stir outside the tech world. The Atlantic argues that Mythos is not an internal Anthropic issue but a societal risk: the model could independently detect thousands of serious vulnerabilities in virtually all common operating systems and browsers. A capability that until recently was reserved only for state-sponsored hackers. The US Treasury Secretary and the Federal Reserve called top executives of the largest banks together for an emergency meeting, fearing a new systemic risk to the financial sector. Gary Marcus nuances: the announcement was made too theatrically and the actual risks are less acute than suggested.
Google serves up three updates in one fell swoop
Gemini gets notebooks that let you bundle chats, files and instructions on a project-by-project basis and synchronize them with NotebookLM. At the same time, the Gemini app can now generate interactive 3D models and customizable graphs, letting you tweak variables or rotate objects, directly in chat. What was vibe coding with Tom a while ago during a lecture is now just baked in and accessible without any technical background. That remains somewhat mindblowing.
The third update is an AI inbox in Gmail that prioritizes and summarizes emails, but for now it is only available for the Google AI Ultra subscription of 274.99 euros a month. On that front, Microsoft Copilot has an edge anyway.
Microsoft Copilot: useful updates, but which Copilot do we really mean?
Word gets new Copilot features for legal and compliance professionals: word-level Track Changes, contextual comments and real-time progress view. Promising additions, though some features like the Frontier preview are still in rollout and not available everywhere yet.
That fits with the pace at which Microsoft is widely integrating AI into its product range: the name Copilot is now popping up in some 80 products and features, indicating how much is in motion at once. For users and IT managers, that requires orientation to keep an overview, and that’s exactly where guidance can make all the difference.
European pressure increases
The European Commission is investigating whether ChatGPT should be classified as a “very large online search engine” under the Digital Services Act. With more than 45 million monthly EU users, OpenAI sits well above the threshold, which if found positive would mean mandatory risk assessments, audits and potentially large fines.
Meanwhile, Microsoft announced that starting April 17, Copilot could process data outside the EU at high load unless administrators actively disable it. Whether that is a real new risk or already covered by existing EU-US agreements is, for now, disputed.
Chatbots answer 12% correctly, yet AI takes over tasks
Research by Emerce shows that chatbots and voice assistants answer only 12% of customer questions correctly. Users also often find it difficult to contact a human employee. At the same time, 20% of full-time working Americans say in a survey by Epoch AI and Ipsos that AI has taken over tasks from them. The nuance that the headline leaves out: these are almost exclusively summarizing, rewriting, information retrieval and preliminary analysis. Tasks, in short, that most employees weren’t awake to anyway.
That brings Tom to a broader observation with which we would like to conclude. At the Dutch Enterprise Chamber, a lawyer once again ended up in the dock after referring to case law that simply doesn’t exist. Supervisors are registering more and more such reports. Critical thinking and proofreading are basic skills that AI is not replacing, but in the AI era may make them more important than ever.
We will be back in two weeks with the next edition.
About Tom Van ‘t veld
Tom has been working for Xylos for years, where he started as a Microsoft Office trainer and grew into the driving force behind innovative learning concepts. He is at the basis of OASE, Xylos’ online learning platform, and developed, among other things, the Digital Coach concept, a Microsoft Teams Escape Room app and the mAindset game that allows employees to playfully learn to promote AI. As a Learning Innovator, he has been increasingly focusing his gaze in recent years on what AI means for the way we learn and work. Respond or chat on? Find him at LinkedIn.
