TechNewsFirst

The piece's whole argument depends on an analogy holding across three very different situations: Amazon hiding a real shipping cost, Facebook disguising an ad as a post, and AI executives voicing genuine safety concern. The first two are documented business practices with internal company data confirming the mechanism, Amazon's own price-comparison logic and Facebook's own engagement-drop testing. The third has no equivalent internal admission; nobody at Anthropic, OpenAI, or Google DeepMind has stated, on the record, that safety messaging is calculated marketing. The piece is honest that this is a pattern it is presenting as exactly that, not a proven case of coordination, but a critic could reasonably say the analogy is doing more persuasive work than the evidence supports: Amazon and Facebook's tricks were provable via leaked internal testing data; the AI-doom-as-marketing claim has no equivalent internal document anywhere in this piece.

On the profitability half, the piece's own caveat, that other reporting could not independently verify Anthropic's figures and Anthropic declined to comment, cuts both ways. It is also possible Anthropic's adjusted operating income framing is standard practice for high-growth, equity-heavy tech companies, since many pre-IPO firms report adjusted metrics and GAAP-versus-adjusted framing is not unique to this industry, in which case the piece's skepticism, however well labeled, risks reading as more damning than a routine accounting convention actually warrants. The weakest single claim is the unlinked Asay quote: without a direct source, a reader can't check how much context surrounded it or whether it was as unequivocal as rendered here.