Bern 2026: putting standards to the field test

On 2 November 2026, Bern hosted the 3rd Swiss Forum on Business and Human Rights, in the year the OECD Guidelines for Multinational Enterprises turn 50 and the UN Guiding Principles on Business and Human Rights turn 15. The programme covered artificial intelligence, finance, the just transition and the role of boards. Discussions were held under the Chatham House Rule, so we share their substance without quoting or attributing any remark to a speaker or an organization. One theme ran through the day, from palm oil to AI: we are not short of principles, we are short of the means to enforce them. This post builds on our account of an earlier forum in Council of Europe and on our work on how smallholder farmers adopt digital tools.

Palm oil: thirty years of standards, the same debates

The opening messages set a clear frame. Due diligence is an integral part of doing business: it is complex, but it serves companies’ own interests. Free trade agreements were presented as a lever to favor cooperative relationships with partners such as India, Thailand, Vietnam, Malaysia and China.

Switzerland has a tool few countries share: the popular vote. The EFTA-Indonesia economic partnership agreement, which grants tariff preferences on palm oil only when sustainability criteria are met, entered into force after voters approved it in March 2021. The EFTA-Malaysia agreement, approved by Parliament in June 2026, now faces a referendum driven by concerns over palm oil, deforestation and labor rights. It was also suggested that Switzerland stands apart by explicitly including the concept of forced labor in a free trade agreement.

One argument struck us in particular. Smallholders do what they can with a crop that Europeans themselves introduced during colonial times. Nothing is black or white, and practices are brought in line with expectations through dialogue, not by cutting economic ties.

We share this rejection of disengagement. As we have argued regarding forced labor import bans, walking away can worsen the situation of the very workers one claims to protect. But dialogue alone is not enough either.

I first set foot on an oil palm plantation in 1998. The problems and debates were exactly those raised in Bern: deforestation, forced labor, the need for local jobs. Nearly thirty years later, they have not gone away; they have changed scale. We have to face it: the standards built over three decades have not solved the problem.

Our diagnosis: these standards lack enforcement. Europe has shifted toward binding rules, with the EU Deforestation Regulation, whose application has been postponed more than once, and the EU Forced Labour Regulation, which applies from December 2027. Supply chains still need the means to comply.

Three levers strike us as decisive:

• training for producers and their organizations, as close to the field as possible;
• access to capital to invest locally in practices and equipment;
• sourcing diversification by international buyers, rather than betting everything on palm oil while hiding behind standards, without giving supply chains the means to apply them where the problems are concrete.

This is the thinking behind our call to secure farm supply in an age of compounding risk: treating farmers as strategic partners, not commodity suppliers.

AI and responsible business: humility over narrative

The first plenary focused on artificial intelligence. There was broad agreement that the topic is urgent, and that three gaps frame the debate.

• A transparency gap between what companies announce and what they actually implement.
• An accountability gap, with close to zero clarity on responsibilities and checks and balances.
• A power gap, driven by concentration among a handful of players, even as the impacts reach democratic infrastructure itself.

The question is no longer whether we want AI, but what kind of AI we want. And this is not a technical problem: it is a societal issue that calls for a holistic response.

AI is now everywhere, raising questions about labor, non-discrimination, freedom of expression and access to information. For fifteen years, the UN Guiding Principles have offered a fit-for-purpose framework. They need to be embedded at every stage, from inception and design to customer use, with a risk-mitigation mindset from day zero. The UN Human Rights Office’s B-Tech Project translates this into practice, with foundational papers on generative AI, including a taxonomy of human rights risks, and a briefing on rights-respecting investment in technology companies. Investors and legal departments alike would benefit from using them, alongside the EU Artificial Intelligence Act, itself already amended in 2026.

Our view is more nuanced. On paper, this approach is beyond dispute. In practice, it remains far from reality: the most widely deployed AI solutions were not built this way. They evolve reactively, driven by market needs and by a growing number of regulatory power struggles.

Technology companies in the room described their approaches, sometimes including teams dedicated to due diligence. That transparency is welcome. But it sidesteps a fundamental question: the leverage a platform exerts over the third-party businesses operating through its application ecosystem. A major AI company never acts alone; it sits within a value chain whose rules it shapes.

These issues call for humility and an acknowledgment of their complexity. They call for genuine stakeholder engagement, rather than a narrative suggesting everything is under control, which everyone knows is impossible given the unknowns that come with AI’s mass adoption. Likewise, relying on the annual review of a broad voluntary initiative to claim third-party assurance is very weak. Such an exercise does not claim to bring the necessary interdisciplinary expertise, nor to assess how AI is vertically integrated across a company’s entire offering.

In a very different sector, we find the palm oil pattern again: a solid normative framework, self-declared implementation, and an enforcement gap that no narrative can close.

From principle to practice: digital tools for smallholders

If enforcement is the missing link, the question becomes: how do we carry due diligence all the way to the first link in the chain, where smallholders operate? Digital tools are part of the answer, provided they avoid the pitfalls seen in AI: tools designed far from their users, and compliance that is displayed rather than lived.

The barriers are well documented. European Commission work on helping smallholders access digital agricultural extension and advisory services points to low digital literacy, a lack of devices and the cost of connectivity. Analysis on making digitalisation work for African smallholders points the same way: technology only creates value when it fits concrete social and economic realities.

We have drawn several lessons from the field, set out in our takeaways on digital inclusion:

• start from the user: a farmer-centric approach, tested in focus groups, that respects constraints on budget, literacy and infrastructure;
• design for frugality: offline mode, text-to-speech for users less comfortable with reading, a simple access link sent by text message;
• start small: one or two features that are genuinely useful to farmers, and limited data collection to ensure quality;
• organize human support: field officers, access points and tailored content, because no tool gets adopted on its own.

This is the approach behind our SUTTI digital solution, which combines blended training, traceability and impact measurement. It speaks directly to the three levers above. Training becomes accessible at scale, in farmers’ own languages. Transaction and production data can open the door to finance. And a detailed understanding of farms helps buyers support diversification rather than suffer it, as we explain in our ten keys to inclusive agricultural supply chain programs.

Digital tools can also strengthen what due diligence systems often lack: the voice of workers and farmers. The Forum itself devoted a workshop to grievance mechanisms that are accessible to people far upstream in value chains. A simple, safe and reliable digital channel can surface weak signals long before an audit would catch them.

One condition remains: data must also flow back to farmers, as advice, more transparent prices or access to credit. Otherwise, the digital tool becomes just another layer of control, one more standard without the means to enforce it.

Thirty years of standards have taught us one thing: compliance cannot be decreed; it is built with those who produce. That holds for palm oil, cocoa and rubber. It holds for AI too.

Image @Ksapa