How is the definition of “ethical AI” shifting as different regions set their own rules?
The definition of ethical AI is moving from a universal ideal to a regional reality. Previously, the conversation centred on principles assumed to be transferable, such as fairness, transparency, accountability, and human oversight. Those still matter, but they no longer travel cleanly across borders. What counts as ethical in one jurisdiction can appear insufficient or non-compliant in another.
The EU has anchored its approach in risk classification and rights protection through the AI Act. The United States has leaned toward sector-led oversight and voluntary commitments. China has prioritised content control and social stability. Each position reflects a different view of what AI is for and who it ultimately answers to.
The numbers also show support for these developments. Our 2026 Confidence Barometer has shown that 92% of executives say they favour a technology partner of the same nationality, and for 47%, it’s a decisive factor. Trust was cited as the biggest reason for this, with over half (57%) linking their concerns directly to security issues. What we’re witnessing is provenance becoming part of the value proposition.
So, the working definition is shifting. Ethical AI is no longer just AI that is in the abstract. It’s AI that earns the right to scale, built with clear accountability, transparent data sourcing, and genuine alignment with the values of the markets it serves, not only the one it was built in.
Is there a growing gap between how much leaders trust AI and how carefully they use it in practice?
Yes, and it is one of the clearest tensions in the current AI moment. Confidence in the technology is running well ahead of the discipline around how it is used.
The data makes the gap hard to ignore. The majority (89%) of leaders surveyed trust AI to guide company decisions, yet 39% admit to sharing confidential data with AI tools they do not fully trust. The same people who are comfortable putting AI at the centre of strategic thinking are also, in practice, handing it information they would not hand to most third parties. We’ve seen this pattern hold across country, company size, and sector, which is telling us something important. It is not a regional quirk or small-company shortcut, it’s a leadership issue.
What is really happening is that adoption has moved faster than internal governance. Tools have entered the workflow through individuals and teams before policies, controls and training caught up.
The responsibility sits squarely with leadership. Just as organisations developed principles for how they use cloud, mobile, or any other category of enterprise technology. AI now needs the same treatment and that means clear standards for what can be shared with which tools, named accountability for AI decisions, and proper review of vendors and data flows. Importantly, none of this slows innovation, it’s what makes innovation safe to scale.
What does the UK’s current regulatory position mean for business confidence on both sides of the Atlantic?
The UK is in an unusual and increasingly useful position. Sitting outside the EU’s prescriptive framework but not inside the US’s lighter-touch model, it is using regulatory distance as a commercial signal. The data suggests the strategy is working. 82% of surveyed UK leaders believe the government is implementing the right policies, up 8 points year on year, and 90% say UK firms are internationally competitive on AI.
The US remains the most enthusiastic at 92%, up 19 points from 2024. France and Spain sit at 81%, Germany at 79%, and Italy at 75%. The UK is moving in the same direction as its European peers, but with a different posture, neither aligned with Brussels nor with Washington.
The middle position shapes how UK executives think about partnerships. Trust is split almost evenly between home (56%) and Europe (53%), which suggests the UK is ideally placed as a bridge between the regions. For businesses on both sides of the Atlantic, that makes the UK a useful proving ground, a place that is regulated enough to be credible and flexible enough to move quickly.
Is the pace of AI investment moving faster than the ethical frameworks being built to support it?
Yes, and the figures show both sides of that picture. 87% of executives plan to increase AI investment. 53% of them significantly, and 94% believe technology can solve the great challenges of our time. A majority (83%) are confident in a sustainable rollout, with only 17% worries about a bubble. That is a striking level of conviction.
The risk is that the optimism becomes complacency, as capital is running ahead of rulebooks and history offers a clear parallel. For instance, we’ve seen social media scale to billions of users before meaningful guardrails existed, and a decade on, society is still grappling with the consequences for mental health, elections, and privacy. The technology overran regulations, and the cleanup has been slower and costlier than the build. AI is on a faster trajectory and wider footprint, which means the cost of repeating the pattern could be even higher.
The encouraging signal is that ethical infrastructure is starting to catch up, particularly in Europe. The partnership between VivaTech and the World Economic Forum to launch the European Centre of AI Excellence in Paris is a useful example. It brings together businesses, policymakers, and researchers to develop and scale AI in a way that explicitly balances innovation with individual rights, the kind of institution that did not exist when social media took off. Its presence now is part of what could make this cycle different.
Investment without scaffolding is the real risk, and scaffolding without investment slows everyone down. The leaders and economies that get this right will be the ones building both at the same time, treating ethical infrastructure as a feature of credible scale.
What do the AI startups leading the next wave have in common?
The winners look less like generalists and more like specialists with discipline. VivaTech’s 2026 Top 100 Rising European Startups points to a clear shared pattern: vertical specialization over generic AI. They are building for specific use cases in content (ElevenLabs, Lovable), productivity (Dust, n8n) and voice (Parloa), while regulated industries like LegalTech (Lawhive), HealthTech (Nabla) and HR Tech (Skello) are now actively embracing AI rather than resisting it.
What ties them together is the combination of technical ambition with enterprise-grade trust-building and cross-border fluency. That is exactly what large buyers are rewarding, and it shows up in the wider numbers too. 92% of executives are confident in maintaining employment levels over the next 12 months, which suggests the most credible AI propositions are being framed as augmentation rather than displacement. The supporting technology stack is also maturing around them: 76% have invested in quantum computing, 45% of them strongly, and 80% in robotic process automation.
The next wave is being led by startups that have stopped trying to be everything and started becoming indispensable somewhere specific.
