Articles such as this include a div inside the main h1; readability seems to pick out the h1 as containing the article title (correctly), but does not strip the div tag. As a result, the article.title returned from readability's .parse() includes that bit of HTML, which appears as plaintext (eg in the browser title bar) when Columnate updates the page to use that "simplified" title.
<h1 class="blog-post-title"><div style="max-width: 480px;">What Is ChatGPT Doing … and Why Does It Work?</div></h1>