Anthropic's choice to anthropomorphize

Starting from the company’s original choice to name its AI product Claude, a clearly human name, Anthropic has consistently taken steps to anthropomorphize its software. I find this increasingly troubling, and worth paying attention to.

While most of the major AI companies go out of their way to not anthropomorphize their products, Anthropic does the opposite. Claude responds to @Claude in Slack, and joins web meetings as itself, encouraging people to perceive it as a teammate instead of computer software. Claude agents attribute authorship to “Claude” at every opportunity, and Anthropic even weaves an invisible (to humans) signature into all text generated with Claude, in the name of “safety” of course.

In ways both subtle and obvious, Anthropic seems to be working hard to create a perception of one Claude, an independent entity, as opposed to various instances of computer software run in service of human operators.