Why AI agents’ success or failure depends on permissions
Start here
AI-assisted source synthesis · not an independent investigation.
On October 28, 2026, PYMNTS reported that Karen Webster believes AI agents may be blocked if they access banks, merchants, and other systems on a “execute first, get approval later” basis due to issues involving risk, costs, and the assignment of responsibility. The article summarizes four approaches for companies: block agents from accessing systems, charge for access, stop building related capabilities, or invite agents to connect based on their own circumstances. Citing Amazon’s blocking of Muse and Walmart’s opening of shopping through Agent Gateway as examples, the report illustrates how different companies have adopted different approaches to agent permissions.
AI-written from reports · updated 2 hours ago
Timeline
Follow the reporting to see the different sides of the story.
- PYMNTSSelectedWhy AI agents' success or failure depends on permissions
Karen Webster argues that if AI agents access banks, merchants and other systems by acting first and obtaining approval later, they will be blocked because of risks, costs and questions over liability. The article outlines four ways companies can respond: block them, charge them, stop building, or invite access on their own terms, illustrated by cases including Amazon blocking Muse and Walmart opening shopping through Agent Gateway.
Heat over time
Not enough consecutive observations yet to draw a trend.