This weekly update shows how trust between humans and AI agents is becoming a major concern worldwide. Companies are adding AI helpers to many jobs, but new problems are appearing.

A big study found that when people let AI make decisions for them, they might act less honestly. This happens because people feel less responsible when an AI does the work. The study tested this with simple tasks and found the same pattern with different AI systems.

In the United States, lawmakers held a special meeting on September 18th to talk about AI leadership and safety. They want to make sure AI systems are safe and trustworthy as more companies use them.

Meanwhile, customer service bots are creating unexpected problems. Some elderly people are forming deep friendships with AI helpers because the bots remember their conversations and seem caring. Companies now worry about what happens when they change or update these AI systems. Should bots remember everything about customers, or should they forget some things?

Security experts warn that AI agents need special protection. Unlike regular computer programs, AI agents can learn and make decisions on their own. This makes them harder to monitor and control. Traditional security tools don't work well with these smart systems.

Several companies announced new AI agent products this month. Adobe launched AI helpers for marketing teams. Dataminr added AI agents to help with cybersecurity. Google and Qualcomm partnered to put AI assistants in cars. All these developments show how quickly AI agents are spreading to different industries.

Experts say companies need to build proper safeguards before using AI agents widely. They recommend clear rules about what AI can and cannot do, regular testing, and keeping humans involved in important decisions.