A paper published on arXiv cs.CY on August 6, 2026, titled "Agentic AI: User Empowerment or Enclosure?" explores the potential societal impact of agentic AI systems. The authors argue that the outcome—user empowerment or enclosure—depends on more than just the technology itself.
Key Points
- Agentic AI systems are designed to act on users' behalf, performing tasks such as content filtering, price negotiation, and service selection.
- The paper conducts a comparative case analysis across four established domains: browser-based ad blockers, platform recommender systems, financial robo-advisors, and email spam governance.
- Decisions regarding whose interests agents serve in these domains were resolved through technical arrangements, including API choices, protocol governance, industry standards, and default configurations.
- These technical arrangements are identified as political decisions, a concept the authors term "depoliticization" within technological systems.
- A consequential effect of this depoliticization is that individual outcomes and collective contestation capacity can diverge.
- An example provided is the improvement in spam inbox quality alongside a collapse in organized capacity to contest spam governance.
Context
According to the authors, the question of whether agentic AI will empower users remains open. They suggest that the answer is not solely determined by technological advancements. The paper draws on political theory to analyze how technical arrangements in past domains have shaped the interests served by agentic systems.
Why It Matters
This analysis is relevant for builders and policymakers developing agentic AI, as it highlights how technical design choices can have political implications, potentially affecting user autonomy and the capacity for collective action.
What To Do
- Read the full paper on arXiv to understand the detailed comparative case analysis.
- Note the specific technical arrangements (API choices, protocol governance, industry standards, default configurations) identified as political decisions.
- Consider how these historical examples might inform the design and governance of new agentic AI systems.
- Watch for further research or policy discussions that address the depoliticization of technical systems in AI development.