In the wild · five-star fakes
AI-written reviews
30-second gist~30s read
AI now writes a meaningful share of online reviews on Amazon, Trustpilot, Google Maps, and the app stores. It's cheap, fast, and easy to scale. Not all of it is malicious — sometimes it's just used to inflate volume — but a lot is paid promotion dressed as ordinary opinion.
The signs are subtle. The reviews are well-spelled, on-topic, polite. They just feel a touch too generic, a touch too uniform, and they show up in suspicious clusters.
If you want more
How to spot one
- The whole feed reads in one voice. Real reviews have rough edges — typos, weird grammar, strong opinions. AI-written reviews are tidy.
- Burst patterns. 30 five-star reviews in three days, then nothing for months.
- Generic praise, no specifics. "Great product, would recommend" with no mention of what was actually used.
- Empty reviewer profiles. No history, no other reviews, the account opened the same week.
- Always positive or always negative. Real reviewers vary; pay-to-spam batches don't.
What platforms are doing
Amazon, Trustpilot, and Google all run detection systems and remove millions of fake reviews a year. Detection is improving, but so is generation. The honest reader's defence is the same as ever: weight reviews less, look for ones with specifics that ring true, ask a friend who has actually used the thing.