I found this piece very hard to parse. I can’t put my finger on why. Almost like the author speaks English and a second language and is constructing the writing based on dictionary/thesaurus definition not common usage.

Like what exactly did you find?

That people can pass the technical side of the interview and fail the culture fit? Not news bro.

That anthropic isn’t interested in what you say you believe but rather in how your beliefs inform your choices and actions? The former is useless (worse than useless because some of the worst behavior is from people who espouse some positive virtues) the latter is what matters.

The only part of this piece I can parse without struggling is the end (and even here why would you use the word alignment?)

“The gap is between alignment-as-stated and alignment-as-demonstrated: the round requires showing where your values have conflicted with something and what you chose, not stating that you hold the right values.”

Anthropic has been describe by more than one person not as a company, but a cult, the cult of Claude.

It makes sense the culture round filters out so many infidels.

This is from our interview research series — we aggregated 15 firsthand accounts from Anthropic ML and software engineering candidates collected in early 2026. The finding that surprised us most: culture round rejections arrive 2 to 3 days after the onsite, while technical failures arrive within 24 hours. That timing difference let us isolate where the process actually ends for most candidates who reach the virtual onsite. The failure pattern in the culture round isn't values mismatch — it's framing. Candidates who gave well-reasoned answers about responsible AI and technical rigor were rejected; the round is looking for Anthropic-specific positions, not professional values. Happy to be corrected by anyone with more recent data.