Some thoughts on the case where a company deleted its production database while using Claude: the direct blame belongs to the person who misconfigured the environment, not to the model itself.
At the same time, we constantly see examples where humans use AI to accomplish something impressive and the credit goes to the AI. Headlines rarely say that a person used AI to solve a problem. They say that AI solved it.
Because of that, it is completely understandable that people also blame the AI when it fails. If labs want the upside of overhyping the technology to seek more funding and profits, they should also expect the public to push back when these systems cause damage.