Users of Meta Muse, xAI Grok, OpenAI Dots, and Instinct reported blocks that prevented their AI agents from performing tasks on major sites, The Verge AI reports.
These are services that are meant to operate autonomously on the internet—for example, booking a table, sorting emails, or purchasing goods. The source does not specify which sites imposed the restrictions or how.
The practical implication of this story is that the capabilities of autonomous agents depend not only on their developers but also on access to third-party web services. There is not enough data yet to gauge the scale of the phenomenon or whether it will become permanent.
editorial commentary
Why it matters
If such restrictions expand, autonomous agents will depend on formal access rules and compatibility with third-party sites. The next observable signals will be statements by site owners or agent developers about the reasons for the blocks and ways to circumvent them. The main uncertainty is the lack of data on the scale and duration of the problem.