Short answer: it does not penalise content for being AI-made. Google evaluates the quality and usefulness of content, not how it was produced. Original, useful content ranks whether a person or an AI wrote it. Thin content built to game the ranking gets penalised, again whether a person or a model wrote it. The test is quality, not origin.
Google's official position is clear: automation, including AI, can produce useful content, and using AI is not spam in itself. What Google fights is content produced primarily to manipulate search rankings rather than help people. Recent core updates reinforce that quality detection, targeting thin, low-value content regardless of its source (Google Search Central).
Google assesses content through the E-E-A-T lens: Experience, Expertise, Authoritativeness and Trustworthiness. This applies equally to human and AI-generated content. And this is exactly where content made purely with AI tends to fail: it can produce correct text, but it has no lived experience, no first-hand cases, no expert judgement. It writes about the work without ever having done it.
In the Content Engine and SEO & AI Search, we use AI for speed, but experience and authority come from real people, which is what Google (and the buyer) values. AI to scale production, experts to guarantee E-E-A-T.
If you are scaling content with AI and watching traffic fall, you are probably on the wrong side of the quality filter. A Diagnosis shows you where.
References - Google Search Central, guidance on AI-generated content