To that point, it seems that he enjoyed interacting with Claude. Sometimes I wonder how many of the bad experiences result from what Jung would have called projections of the user's self onto the AI through their prompts.
This is the most ridiculous variation of "skill-issue" or "user error" I have heard so far. Give back the $0.50 in shill money to Anthropic.
I don't take those statements at face value - he's employed by them for a lot of money and is being given access to high level (and expensive) tools. What he put out was effectively a press release, and I don't doubt that a bunch of marketing, PR, and lawyers reviewed it before it was allowed to be posted.