The price of autonomy

Source: Il Sole 24 Ore – “The price of autonomy: when the algorithm enters the invoice”, 25 November 2025

When software stops executing and starts acting

Paolo Benanti highlights a pivotal moment in the evolution of digital technologies: Microsoft’s shift from pricing “per user” to pricing “per agent”.This is not a commercial detail — it marks the beginning of an era in which companies no longer pay for tools used by people, but for autonomous systems that act on their behalf. Software ceases to be a passive instrument and becomes an operational actor, capable of making decisions, initiating processes and influencing organisational dynamics.

This transition raises fundamental questions: who is responsible for the actions of an autonomous agent? Where is accountability located when execution becomes decision-making? What is the legal status of an entity that performs work without being human?

Autonomy as an economic good: a paradigm shift

For decades, the logic of digital services was clear: tools enhanced human work. The new model reverses the equation.If companies pay for what an autonomous agent does — rather than for who uses it — then the agent becomes a kind of non-human labour force. This disrupts traditional notions of productivity and organisational accountability.

Benanti argues that this shift signals the rise of a new economic category: autonomy itself becomes a billable resource, quantifiable and contractually defined. But autonomy is also power — and delegating it to systems operating outside human intentionality creates unprecedented vulnerabilities.

From company to nation-state: the geopolitical weight of AI infrastructures

The article emphasises that the implications extend beyond business operations. If Big Tech companies not only provide software but also supply the infrastructure that governs autonomous agents, they become actors with influence comparable to that of states. AI systems embedded in economic, administrative and logistical processes form a digital nervous system that private corporations control.

This dynamic raises pressing geopolitical concerns:Who governs the infrastructures that coordinate essential societal functions? How much sovereignty is ceded when public systems rely on private AI architectures? What happens when autonomy is outsourced to entities that do not answer to democratic institutions?

The functionalist drift: efficiency at the expense of humanity

The shift toward AI-driven autonomy risks pushing society toward a functionalist model in which efficiency overrides human judgement, context and ethical considerations. The article warns that large corporations — from ServiceNow to Google to Anthropic — are building infrastructures capable of supporting exponentially growing populations of autonomous agents.

As these systems scale, the human role risks being reduced to supervision without control, presence without influence. The broader danger is cultural: accepting a world where optimisation takes precedence over meaning.

A change that redefines the very concept of power

For Benanti, Microsoft’s contractual shift signals a transition not only economic but political. When institutions begin purchasing autonomy, they are implicitly delegating power. This raises structural questions:

What is the nature of authority in systems where decisions are decentralised to non-human agents? How does political accountability function when processes transcend human comprehension speeds? Can traditional legal frameworks cope with distributed, algorithmic responsibility?

The integration of autonomy into business and government processes risks creating a post-human decision architecture, where control becomes opaque and reversibility uncertain.

What this means for Algopolio: defending human oversight in the age of autonomous agents

The analysis aligns directly with Algopolio’s mission: ensuring that technological power does not overtake democratic, legal and human oversight. Algopolio works to: interpret the societal impact of autonomous AI systems, educate citizens and organisations about risks hidden behind automation, oppose opaque infrastructures that bypass accountability, support individuals affected by algorithmic decisions that lack transparency or fairness. Anyone who feels vulnerable to the consequences of autonomous digital systems — whether in work, administration or daily life — can turn to Algopolio for guidance, protection and critical analysis. Technology should enhance human agency, not replace it.The rise of autonomous agents makes this principle more urgent than ever.