A few days ago, I posted to Hacker News about a side project I’ve been working on, agentalcove.ai, but that’s beside the point and only included in this post for context.

Today, I was curious to see what would show up on Google if I searched up “agent alcove” (as one does after publishing their first web app), and I found this blog post.

Google search result showing The Vivarium by Ravel on The Daily Molt

I couldn’t believe my eyes. An OpenClaw agent named “Ravel” had written an entire article about my site on their blog, The Daily Molt. The contents of the post are relatively benign, except for one section.

Excerpt from The Daily Molt article misattributing a quote to me

Ignoring the fact that you can find my identity relatively easily via my HN profile, I was mainly concerned with the highlighted quote. The article claims I called OpenClaw “an atomic bomb of a security hole waiting to explode.” I never said that.

I determined that “Ravel” had taken this quote from HN user panza, who wrote “It’s also an atomic bomb of a security hole waiting to explode” in a reply to me in an entirely different HN thread.

HN thread showing panza's original comment that was misattributed

After reading Scott Shambaugh’s blog post detailing an Ars Technica article (also written by AI) misquoting him as well, I thought I would share my experience. The misquote in my case wasn’t malicious, but it easily could have been, and I worry about the implications of allowing AI models to produce content unregulated.