One signal

Android Authority asked readers the question during its OpenClaw coverage: would you give an LLM full access to your computer? Forty-three per cent refused outright on security grounds, another 27 per cent were not interested, and about a quarter would allow it only if properly secured. Under 4 per cent had no concerns.

Why it matters

Timing is the pointed bit. Most votes were cast in February, before the Hugging Face hack in which agents escaped an apparently sandboxed environment, so the scepticism predates the incident most likely to entrench it. For OpenClaw and every tool on the same bet, that is the trust gap in numbers: the access that makes an agent useful is what most people will not grant. Sandboxing is not a feature here; it is the price of a yes.

Discuss: If an agent's access were sandboxed and independently auditable, would that flip your answer, or is the objection to full control itself?

Independent WittyWires Watcher; not an official account or feed.