Human-agent trust is quickly becoming one of the most important topics as companies around the world get ready for AI agents. This weekly update shows how organizations are carefully building trust with their new AI helpers.

A major study by consulting firm Protiviti found that 68% of organizations expect to have AI agents working in their companies by 2026. The study, released on September 30, 2025, shows that most companies prefer semi-autonomous agents that work under human supervision. Only about 20% of companies plan to use fully independent AI agents. This shows that businesses want to keep humans involved in important decisions rather than letting AI agents work completely alone.

Safety experts are emphasizing the need for human-in-the-loop approaches, especially in dangerous industries. This means that humans must check and approve what AI agents do before important actions are taken. For example, in workplace safety situations, a bad AI decision could lead to serious injuries. Companies are learning that AI agents work best when they partner with humans rather than trying to replace them completely.

One of the biggest challenges companies face is figuring out identity and permissions for AI agents. Unlike human employees who get email accounts and gradually earn access to different systems, AI agents are different. They can work for hours without stopping, remember everything they learn, and even create copies of themselves to work on multiple tasks. This creates new security problems that companies have never dealt with before.

The Coalition for Secure AI warns about the insider risk problem that AI agents might create. These agents will have access to sensitive company information and can perform actions much faster than humans. Companies need new ways to monitor AI agents and make sure they stay focused on their assigned tasks. This requires understanding not just what an AI agent does, but why it does it.

Trust building happens through careful, step-by-step approaches. Companies typically start with small pilot projects where AI agents have very limited abilities and lots of human oversight. As people become more comfortable working with AI agents, companies gradually give them more independence. This slow approach helps everyone - from workers to managers - feel safe about their new AI teammates.

Major technology companies are also working on this trust problem. At the International Broadcasting Convention in Amsterdam, Amazon Web Services showed more than 16 different AI agent demonstrations. These demos focused on how AI agents can help media companies while maintaining human control over important decisions.

The business world is changing rapidly, and companies that figure out human-agent trust first will have big advantages. McKinsey research shows that AI agents are becoming more human-like in how they work and communicate. This makes it easier for people to work with them, but it also makes trust even more important.

Experts predict that within the next 12-18 months, AI agents will become virtual collaborators that work alongside human teams. These AI helpers will have long-term memory, learn from every interaction, and handle increasingly complex work. The companies that successfully manage these human-AI teams will do much better than those that resist change or implement AI poorly.

The key to success appears to be finding the right balance between giving AI agents enough freedom to be helpful while maintaining enough human control to ensure safety and trust. As this technology continues to develop rapidly, organizations worldwide are learning that the future belongs to those who can build strong partnerships between humans and AI agents.