-
It is not an overreaction at all. What you describe is a clear sign of poor QA and a weak content process. Any good SEO agency would treat your existing content as an asset. They would protect brand tone, product relevance and factual accuracy. What they delivered shows they are running bulk AI prompts with little or no review.
Here is what is going wrong from an SEO point of view:
Removing product context harms topical authority as if an article ranks because it is linked to your product and your niche, stripping that out can weaken relevance. Google reads your pages in context of your whole site. Going generic can dilute that.Recommending a competitor is a huge red flag as no strong workflow should allow that through. It shows no one is doing a final read. Even basic editors would catch this.
The biology references show they are using one prompt template for many clients with no guardrails. Google does not punish AI itself, but people will punish poor quality content that is wrong, irrelevant, misleading or unhelpful. They will pogo stick to the next result below yours… This type of mistake is a trust killer.
AI is fine when it is well edited and provides real value. It fails when it is lazy, hallucinated or stuffed with filler. What you describe is the type of content update that can cause traffic drops.
More words are not an SEO tactic, Google rewards clarity and usefulness. You get no gain from padding an article. In many cases you hurt the page because users bounce.
Your fear about scale is valid as If they do 30 of these a month, you can end up with a site wide quality issue that is then time consuming to fix. Site quality is judged at the domain level. A flood of poor updates can drag down sections of your site that already perform well, either by reducing conversions or people pogo sticking.
What a good agency would do instead is to keep your tone of voice, add missing sections based on search intent, improve structure, update facts, strengthen internal linking, remove fluff, not add it, only use AI as a support tool, not as the final output.
You are right to question this. Fewer high quality optimisations are always better than many low quality rewrites.
If this is the first batch, I would stop them now and ask for a new workflow. If they cannot explain their process in plain language and show human QA steps, I would move on.
