Reports of people losing data and other resources due to unintended actions from autonomous agents come out practically every week. I don't think it's dishonest to say that could have catastrophic impact on the product/service they're developing.