Sweden's consumer watchdog orders Allente to prove what its sales calls actually say

What happened

On 12 December 2024, Konsumentverket, Sweden's consumer ombudsman, issued an enforcement order against Allente Sverige AB ^[1]. The order requires Allente to clearly inform consumers during telemarketing calls that it is Allente calling and that the purpose of the call is to sell subscriptions.

The ruling followed a complaint that consumers had been misled about who was calling and why. When a consumer picks up the phone and hears a sales pitch, they are entitled to know straightaway who is on the line and what the call is about. That might sound obvious, but in practice it is easy to leave unsaid.

Konsumentverket found that Allente's phone sales did not make those two things clear enough. The order is a reminder that marketing law doesn't just regulate what you say—it regulates what you don't say as well.

What the law requires

Under Chapter 3, section 6 of the Swedish Marketing Act (marknadsföringslagen) ^[2], a commercial message must make clear that it is a commercial message and who is sending it. Silence on those points can itself be misleading.

It's a bit like someone knocking on your door without saying why they're there. Even if they're polite and never lie, you still don't know whether it's a neighbour, a courier or a salesperson until they tell you. Swedish consumer law says the seller has to speak first.

For telemarketing, that means stating the company's name and the fact that the purpose is sales, early in the call and in plain language. It's one of the small details that, when omitted, can turn an ordinary sales pitch into something the law sees as unfair.

The evidence question

The interesting part of this case is not what Allente did or didn't do—it's the question it raises for every company that sells by phone, chat or AI.

If a consumer complains that your sales team or chatbot didn't identify itself properly, can you show what was actually said? If a supervisor asks whether your AI disclosed that it was an AI, do you have a record, or just the consumer's word against yours?

Most companies keep call recordings for quality purposes. Fewer can show that a recording hasn't been edited since the call ended, and hardly any seal the record at the moment the conversation happens. When a dispute arises months later, the lack of a trustworthy record is a weakness.

Konsumentverket's order to Allente doesn't say how the company should prove compliance going forward. It just says what must be communicated. But the gap between a rule and the ability to show you followed it is where disputes live.

How a digital witness fits here

Here, the thing to seal is each call record or chatbot conversation as it ends: the timestamp, the channel (phone, chat or AI), the script version if there is one, and a transcript or summary if the system generates one. Sealing takes a fingerprint of the record, not the content, so the conversation itself never leaves your systems.

When a call happens, the sales platform or chatbot service sends that fingerprint to WitniumChain and gets a witness ID back. If the call is questioned later, anyone you share the record with—whether it's Konsumentverket, an internal auditor or the consumer—can check that it's the one sealed at the time. Change a single word in the transcript and the fingerprint won't match.

The witness doesn't show that the caller followed the script, or that the conversation was fair. It just shows that the record is the one created at the time of the call, unchanged since. That's a narrow thing, but in a dispute it can be the difference between a short conversation and a long one.

You can seal call records with the REST API if your telephony system supports webhooks, or with the SDK if you run your own chat or agent service. For AI chatbots, the MCP server lets agents seal conversations as they happen. The content stays on your systems. The first 2,000 witnesses a month are free. Read the docs.

How it would work

1. Your sales system seals a fingerprint of each call or chat record as the conversation ends.
2. The system stores the witness ID alongside the record.
3. If the record is questioned, anyone you share it with can check that it hasn't changed since it was sealed.

What this means for you

If you sell to consumers by phone, chat or AI, ask yourself:

• Does every script or AI prompt clearly state the company's name and the purpose of the contact?
• Can you show what was actually said on a given call or in a given chat, six months later?
• If a consumer or a supervisor questions a call, do you have a record you can trust, or just the consumer's screenshot?

The law is clear on what must be communicated. The harder question is whether you can demonstrate that it was.

What to watch

Under Article 50 of the AI Act, people interacting with an AI system must as a rule be informed that they are doing so, unless it is obvious from the context ^[3]. For AI-powered customer service in the EU, that disclosure duty has applied since 2 August 2024 for emotion recognition and biometric categorisation systems, and will apply from 2 August 2026 for most other AI systems.

If your chatbot or voice assistant sells to consumers in Sweden or elsewhere in the EU, the combination of the Marketing Act and Article 50 means two things must be clear from the start: that it's an AI, and who is using it to sell.

For enforcements of the AI Act's transparency rules, watch the AI Office's compliance tracker and national supervisors' announcements.

Sources

1. Konsumentverket, "Föreläggande mot Allente Sverige AB", 12 December 2024, https://www.konsumentverket.se/forelaggande/forelaggande-mot-allente-sverige-ab/
2. Marknadsföringslag (2008:486), Chapter 3, section 6, https://www.riksdagen.se/sv/dokument-och-lagar/dokument/svensk-forfattningssamling/marknadsforingslag-2008486_sfs-2008-486/
3. Regulation (EU) 2024/1689 (AI Act), Article 50, https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50