Monthly commentary for July and August 2026

July 2026

For us, July was a month of preparation. The work went into the structure behind the publication: which areas we will explain, in what order, and how each article can show where the facts come from. It is slow work, and it does not show from the outside until the texts start to arrive. We still chose to spend the time there, for a simple reason: a library built carelessly costs more to repair than it costs to build it right from the start.

Outside our own work, all the more happened. On 8 July 2026 the EU's simplification act for AI was signed, formally Regulation (EU) 2026/1744, often called the Digital Omnibus on AI. It was published in the Official Journal of the European Union on 24 July and entered into force on 27 July 2026.

What it did is worth understanding, because it says something about how new this technology still is. The act pushed back the heaviest requirements in the AI Act, those that apply to high-risk AI systems. The rules for such systems in areas such as education, employment and law enforcement start to apply on 2 December 2027. For AI built into products that already have their own safety rules, for example medical devices and lifts, the date is 2 August 2028. The reason given in the act is concrete: the technical standards and national supervision were not ready in time.

Sweden had already taken its step. On 12 June 2026 the government gave five authorities the task of acting as national competent authorities under the AI Act: the Swedish Post and Telecom Authority (Post- och telestyrelsen), the Swedish Authority for Privacy Protection (Integritetsskyddsmyndigheten), the Swedish Financial Supervisory Authority (Finansinspektionen), the Swedish Medical Products Agency (Läkemedelsverket) and Swedac. The assignment runs until 31 December 2026, pending supplementary Swedish legislation.

August 2026, the position as of 25 August

On 2 August 2026 the AI Act's transparency rules began to apply. They are the first rules that an ordinary person notices in everyday life, and they are simple to summarise.

You must be told when you are talking to an AI system rather than to a human. Content created or manipulated by AI must be labelled, both so that a machine can read the label and, in some cases, so that you can see it with your own eyes. This applies, among other things, to images, audio and video that look like real people or events. It also applies to text published on matters of public interest without a human having reviewed it. Anyone who uses tools that claim to read emotions or sort people by biometric traits must say so.

The fines are not symbolic: up to 15 million euros or 3% of global annual turnover, whichever is higher. There is one exception. Generative AI systems that were already on the market before 2 August 2026 have until 2 December 2026 to meet the requirement for machine-readable labelling.

On our side, the library opened in August. The first article explains what AI is, from the ground up and with no prior knowledge needed, and it is available to read on the site. The newsletter is open to anyone who wants new texts straight away. The work continues in the five areas we have set out to explain: technology, demographics, urbanisation, climate and resources, and growing wealth.

The number of unitholders in AIX Dynamic was 22 as of 31 July 2026.

What this commentary does not contain

It contains no figures on performance, no forecasts and no comparisons with the market. The fund is newly launched, and we do not describe performance in our marketing. What we can do instead is explain the technology and the rules that actually shape the areas we write about, and show where every piece of information comes from.

Education, not advice.

This is marketing. Fund units can both rise and fall in value. Read the key information document (KID) and the prospectus before you invest; you will find them on the fund's page at Avanza.