# Weekly Update: Human-Agent Trust

## The Future of AI Agents Is Here

This week brought important news about how AI agents—computer programs that can make decisions and take actions on their own—are changing how we work and live. Experts from all around the world are paying close attention to this change. A big survey of technology leaders found that about 82% of them believe AI agents will play a much bigger role in our lives and in government work within the next 10 years or even sooner. This means AI agents will soon help doctors, teachers, and government workers make important decisions every single day.

## A Big Trust Problem With AI Agents

However, new research this week showed that companies have a serious problem with AI agent trust. Scientists asked 285 security and technology workers about how they keep AI agents safe and under control. What they discovered was shocking: only 18% of these security leaders feel confident that their company's security systems can properly manage AI agents. This means 4 out of 5 security leaders are worried and nervous about whether their AI agents are being used correctly.

The problem is even bigger than most people think. Many companies are using very old and unsafe methods to control AI agents. For example, 44% of companies use old-fashioned API keys (like a digital password) to let AI agents access important systems. Another 43% use simple username and password combinations, and 35% let multiple AI agents share the same login account. All of these methods are unsafe because they are from an older time when computers worked differently. When AI agents use these old methods, they can access sensitive information 24 hours a day without anyone watching.

Another big problem is that most companies cannot see what their AI agents are actually doing. Only 28% of companies can actually trace what an AI agent did back to the person who told it to do that task. Even worse, only 21% of companies keep a real-time list of all their active AI agents working right now. This means that nearly 80% of companies cannot tell you right now what their AI agents are doing or who is responsible if something goes wrong.

## The Good News: AI Agents and Humans Working Together

Despite these challenges, there is great news coming from research on how humans and AI agents can work together successfully. When humans and AI agents collaborate properly, something wonderful happens. Studies show that human performance can improve by up to 29% when people work closely and repeatedly with AI agents. This means people can work faster, smarter, and better when they have AI teammates they trust.

Government agencies are already seeing this success happen in real life. In New Jersey, United States, government employees tested AI agents in safe practice environments before using them in real work. The results were excellent: self-resolved work went up by 50%, response times dropped by 35%, and more than 80% of employees said the AI tools made their work better. These examples show that when AI agents are designed and controlled the right way, they can help people do amazing work.

## Why Trust and Clear Explanations Matter

Another important finding this week came from a study about what older adults need from AI agents. Researchers at Georgia Tech in the United States interviewed older people who use voice AI assistants like Amazon's Alexa. They discovered something crucial: older adults only trust AI agents when they explain their answers clearly. When an AI agent just says "I am 92% sure about this," older people don't trust it. They want to know what information the AI agent is basing its answer on.

This finding is important for everyone, not just older people. AI agents need to show their sources and reasoning if people are going to trust them. When AI agents are like a "black box" that just gives answers without explaining them, people get nervous and worried. But when AI agents work like trustworthy teammates that explain their thinking, people feel comfortable working with them.

## What This Means for the Future

This week's news shows us that we are at an important moment for AI agents. Companies and governments are beginning to use AI agents to help workers make decisions and get work done faster. But they need to solve the trust and security problems first. Leaders need to design new ways to keep AI agents safe, make sure people can see what they are doing, and require them to explain their answers clearly.

The future of work will include humans and AI agents working together as a team. This can be amazing if we build trust the right way. By 2028, it is possible that 15% of all daily work decisions will be made by AI agents or influenced by them. That is why it is so important that we build strong trust between humans and their AI teammates right now.