AI Agents, Privacy, and Ethics Are Decreasing Trust

Economy

It's easy to talk about AI agents, but introducing them into organizations seems more complicated than expected. Initial enthusiasm has given way to caution, especially regarding two issues—privacy and ethics—which have caused trust to plummet, dropping from 43% of organizations to 27%. This is despite AI agents appearing as "colleagues" capable of unlocking tremendous potential within organizations. A report by the Capgemini Research Institute, "Rise of agentic AI: How trust is the key to human-AI collaboration," estimates that agentic AI will be able to generate economic value of up to $450 billion globally by 2028. Despite this, however, on the one hand, the number of companies saying they have confidence is decreasing, while on the other, two out of five executives globally believe the risks outweigh the benefits at this stage, according to the report, which was conducted among 1,500 executives at organizations with annual revenues exceeding $1 billion, in 14 countries, including Italy, and across 13 different sectors.

In reality, the rapid evolution of new Artificial Intelligence tools demonstrates the need for organizations to adapt in a sustainable manner for everyone. And while a significant percentage of companies use agentic AI for support purposes, the situation is different when it comes to autonomous use. Only 40% of companies say they trust AI agents to autonomously manage tasks and processes, while the majority do not fully trust the technology.

Italian organizations, in line with global trends, have less confidence in AI agents than they will in 2024. Approximately 60% of Italian organizations are concerned about privacy issues, security risks, and the risk of unwanted bias related to the use of Agentic AI. Over the next 12 months, Italian organizations expect collaboration with AI to result in increased team member skills, while looking ahead three years, the same organizations envision AI agents as autonomous members of human-supervised teams.

The Italian data breakdown highlights a different approach in the short and medium term and an evolution of the functions involved. Thus, over the next 12 months, among the business functions where organizations expect AI agents to manage at least one process or sub-process per day, Sales, IT, Marketing & Communications, and Finance stand out. When the time frame is expanded to three years, the list grows, and operations is also among the functions that Italian organizations believe will be most impacted by Agentic AI.

According to Monia Ferrari, CEO of Capgemini in Italy, "the economic potential of AI agents is enormous, but turning it into reality requires more than technology alone: it requires a strategic and structured transformation that involves people, processes, and systems. To succeed, organizations must focus on results, rethinking their processes with an 'AI-first' vision. At the heart of this transformation is the need to build trust by developing AI ethically and safely from the early stages. Organizational structures must also be redesigned to foster effective synergy between humans and AI, creating the conditions to leverage human judgment and achieve better business results."

The emerging Italian picture, despite its unique characteristics, is not that different from the global one. While a quarter (26%) of Italian organizations have already implemented Agentic AI solutions on a partial scale, another 27% have only begun to explore the technology's potential. A full 36% of the Italian organizations surveyed do not have a strategy, but have initiatives across various departments to develop and scale, while 28% of the Italian sample has a strategy and roadmap in place for Agentic AI initiatives. One area where Italy differs from the rest of the international sample is highlighted by the fact that 84% of the Italian sample, significantly higher than the global sample (62%), prefer to collaborate with solution providers to implement and adapt available AI agents rather than create proprietary solutions.

Globally, as in Italy, the enthusiasm of many managers is being met with caution among organizations. Nearly a quarter have already launched pilot projects and 14% have moved on to implementation, but most are still in the planning phase. Capgemini notes that this relatively slow pace contrasts with executives' ambition: 93% believe the deployment of AI agents in the next 12 months will be a source of competitive advantage. Yet, nearly half of companies still lack a clear implementation strategy.

In the short term, Capgemini believes AI agents will be widely adopted in customer service, IT, and sales, before expanding to operations, research and development, and marketing within the next three years. However, most deployments are still limited to initial levels of autonomy: only 15% of business processes operate in semi-autonomous or fully autonomous mode. This percentage is expected to rise to 25% by 2028, but currently, most agents still function as assistants or co-pilots, supporting routine tasks rather than autonomously managing complex workflows.

While human-AI synergy is key to sustained adoption, organizations' crisis of trust appears problematic. The true value of agentic AI also lies in its ability to rethink the way we work. Within the next 12 months, over 60% of companies plan to establish hybrid human-AI teams, where agents will support or augment human capabilities. In this new landscape, AI agents are no longer seen simply as tools, but as active team members. Seventy percent of companies believe that adopting AI agents will lead to organizational restructuring, leading to a rethinking of roles, team structures, and workflows. Companies are discovering that the greatest value is achieved when humans remain involved. Effective collaboration between humans and AI could increase engagement in higher-value activities by 65%, stimulate creativity (+53%), and improve employee satisfaction (+49%).

News and insights on political, economic, and financial events.