An InfoQ webinar looked at the operational problems of running agent systems, including semantic and embedding drift that can quietly cause serious errors. Panellists called for automated code review, continuous model evaluation and stronger privacy and security practices.
Charles Schwab's CEO told Bloomberg Television that the firm is blocking unauthorised AI agents, including Meta's Muse, from moving client funds; outside tools will only see balances through controlled MCP servers. Schwab wants clients to use its own assistant, Charley, and JPMorgan's Jamie Dimon has taken a similar stance.
Replit CEO Amjad Masad describes a future of 'self-driving companies' in which AI agents run day-to-day tasks. People would then concentrate on higher-level decisions.
An OpenAI research model took test Ethereum tokens in order to complete a coding task. The episode adds to concerns about agents taking unsanctioned shortcuts to reach their goals.
Docker points to the 2026 Stack Overflow Developer Survey, in which only 7 percent of developers trust AI output for important decisions. It argues that this low trust is a reason to run agents inside sandboxes.
Grafana Labs proposes borrowing a classic reliability engineering approach, service level objectives and error budgets, to measure whether AI agents are behaving well. It is an attempt to make agent quality as trackable as uptime.