The EU AI Act is complex, but your path to AI compliance does not have to be. Here the Klaus team explains deadlines, risk categories and obligations in plain language: concrete, without legalese, with to-dos you can act on.
Since 29 July 2026, Germany has had an AI supervisory authority. As a rule, that is the Federal Network Agency (Bundesnetzagentur). What it can check today, what happens after a request for information or a complaint, which fines are realistic for small businesses and why you do not have to register anything.
Read article
Law firms may use AI, but under stricter rules than most industries. What lawyers’ duty of confidentiality, Section 43e of the Federal Lawyers’ Act (BRAO) and Section 203 of the German Criminal Code (StGB) require when you use ChatGPT, Copilot or Claude, why a data processing agreement is not enough and what applies under the AI Act since the Digital Omnibus.
Read more →The transparency obligations of the EU AI Act have been in effect since 2 August 2026. The good news: most AI-generated texts from small businesses do not need to be labelled. Here is where the obligation really applies, how to label correctly and what the supervisory authority expects.
Read more →Copilot is built into Windows, Edge, Outlook and Teams. Whether company data is safe there depends on which account your team signs in with and how your Microsoft 365 is configured. Here is what applies after the renaming in summer 2026 and which settings you should check now.
Read more →The Digital Omnibus has been in force since 27 July 2026 and amends more than 40 articles of the EU AI Act: postponed high-risk deadlines, a softened AI literacy obligation, relief for small companies. Time for a sober assessment of what this means for you and which obligations started unchanged on 2 August 2026.
Read more →Almost one in two employees uses AI at work, but only a quarter of companies have rules for it. This gap is a real risk for data protection, trade secrets and liability. We show you what belongs in an AI policy and how to set one up in a few days.
Read more →Since February 2025, the EU AI Act has required measures for AI literacy in your team. In July 2026 the obligation was softened, but not abolished, and from August the supervisory authority is watching. What you really need to train and how to prove it.
Read more →In four out of ten companies, employees use private AI tools, often without their bosses knowing. What sounds like initiative is a very real risk for data protection, trade secrets and IT security. Here is how to bring shadow AI into the light without slowing your team down.
Read more →The EU AI Act does not apply all at once, but in stages. And since July 2026, there has been a twist: the Digital Omnibus has amended the timeline after the fact. We show you which obligations already apply, what has applied since 2 August 2026, what is added on 2 December 2026 and where you have until the end of 2027.
Read more →ChatGPT has long been part of everyday life in most offices, often through personal accounts. That is exactly the problem: training on your inputs, no contract, US servers. Here is which plan and which settings let you use ChatGPT properly in your business.
Read more →Claude reads long documents, writes texts and helps with programming. Whether the tool may also see company data depends on how you access it: when it comes to data protection, a personal account and a business plan are worlds apart. Here is how to use Claude properly in your business.
Read more →Prohibited, high-risk, limited or minimal? The EU AI Act sorts every AI application into one of four risk categories. That determines which obligations you face. Here is how to find out where your AI stands.
Read more →