Editorial illustration for EU sees agentic AI as next chapter to deepen use and boost productivity
EU Workers Hand Tasks to AI Agents, Boosting Productivity
EU sees agentic AI as next chapter to deepen use and boost productivity
OpenAI's latest Economic Research report tracks a shift already underway inside European companies: workers are handing tasks to AI agents instead of just asking questions. Among ChatGPT Enterprise customers in the EU, Codex and ChatGPT Work, OpenAI's agentic products, now account for more than half of all output tokens generated. Small and medium businesses are moving in the same direction. Since May 2026, weekly active users of these tools among EU small and medium businesses have grown more than eightfold, with increases recorded in every member state.
That growth isn't even across the bloc. OpenAI's data shows wide gaps between countries in how deeply agentic tools have taken hold, and in what kinds of work employees actually bring to ChatGPT. The report also draws a line between adoption rates and adoption intensity, pointing to a gap between the EU's most AI-active firms and the rest of the pack, and between those EU frontrunners and firms operating at the global frontier. The numbers below lay out just how wide those gaps are, starting with the range between the EU's slowest and fastest adopters.
The takeaway is that EU frontier firms use AI much more intensively than typical EU firms, but still trail firms at the global frontier. In September, frontier firms in the EU generated more than four times as many output tokens per active user as typical EU firms.
Why this matters
OpenAI's framing conveniently arrives attached to a report OpenAI itself produced, which is worth sitting with for a second. The EU has spent two years building out the AI Act, a regulatory structure built largely around chatbot-era risks: hallucination, transparency, data use. Agentic AI, where a tool plans and executes multi-step tasks on its own, is a different animal, and the gap between "broader adoption" and "deeper use" that the quote gestures at is really a gap in oversight, not just skill.
For founders building on top of these models, the opportunity is real: enterprise customers moving from Q&A to task execution means bigger contracts and stickier products. But for researchers and anyone tracking governance, the interesting question isn't whether EU businesses adopt agents faster. It's whether the EU's existing rulebook, written before agents went mainstream, can actually catch what agents do once they're let loose on real workflows.
Watch how regulators respond once agentic tools start making consequential decisions, not just answering questions.
Related Reading
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google to End Free Access to Gemini AI for Non-Subscribers
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
From Archives Neural network builds learning machinery from simple brain rules
From Archives Disney Offers USD 20M for Character.AI Chatbot Tech After Child Harm Claims
Common Questions Answered
What is driving the shift toward agentic AI adoption among EU companies?
European companies are increasingly handing tasks to AI agents rather than simply asking questions, with agentic products like Codex and ChatGPT Work now accounting for more than half of all output tokens generated among ChatGPT Enterprise customers in the EU. This shift represents a move from basic question-answering to deeper, more intensive AI use where tools plan and execute multi-step tasks autonomously.
How do frontier firms in the EU compare to typical EU firms in AI token generation?
According to OpenAI's Economic Research report, frontier firms in the EU generated more than four times as many output tokens per active user as typical EU firms in September. This significant gap demonstrates that while frontier firms use AI much more intensively, they still trail firms at the global frontier in overall AI adoption and productivity gains.
Why is the EU's AI Act potentially inadequate for regulating agentic AI?
The EU's AI Act was built largely around chatbot-era risks such as hallucination, transparency, and data use, but agentic AI represents a fundamentally different challenge where tools autonomously plan and execute multi-step tasks. The regulatory framework may not adequately address the unique risks and governance needs that arise from this deeper, more autonomous form of AI deployment.
What growth trend is OpenAI observing among EU small and medium businesses using agentic tools?
Since May 2026, weekly active users of agentic products among EU small and medium businesses have shown significant growth, indicating that adoption of these autonomous AI tools is expanding beyond large enterprises. This trend suggests that smaller organizations are increasingly recognizing the productivity benefits of delegating complex tasks to AI agents.
Further Reading
- The EU’s next chapter in AI is agentic - OpenAI
- Gemini at Work 2026: Introducing Gemini agent - Google Cloud
- The Transatlantic AI Divide: Why Europe is Falling Behind in the Race for Autonomous Agents - Techstrong.ai
- AI in Europe: From Ambition to Execution in 2026 - World Wide Technology
- AI agents and enterprise transformation: Turning hype into measurable value in 2026 - IT Brief UK
