Star Ratings Are Broken and the AI Shopping Agents Replacing Them Are Not Trusted Either
82% of consumers encounter fake reviews at least once in 12 months. Fake reviews cost global consumers $770.7 billion in bad purchases in 2025. Platforms remove 6.9% of the reviews identified as suspicious. Now AI shopping agents are being positioned as the replacement — and 27% of consumers trust no organisation to operate one on their behalf.
Who has this problem
Shoppers using AI chat assistants for purchase recommendations
People relying on ChatGPT or Perplexity to shop, with no way to audit why a specific product was recommended.
Brands whose products depend on being surfaced by an AI shopping agent
Sellers with no visibility into why an AI recommended a competitor instead of them, unlike the old, at-least-legible star-rating system.
Review platforms and marketplaces
Companies like Trustpilot and Bazaarvoice adapting to AI shopping agents.
Regulators watching AI shopping agents
Consumer-protection bodies weighing how AI recommendations should be disclosed.
Signal timeline
Researchers at the Chinese University of Hong Kong publish FORGE, a benchmark testing whether AI shopping agents can be tricked into recommending a fake brand by swapping the brand name on a real product page; across 12 commercial and open-weight LLMs (including Anthropic, Google, and OpenAI models), a single polluted page fooled agents up to 27% of the time, rising to 73.8% once the top 3 results were replaced.
Keep reading with access
You've seen the scores, who has this problem, and the latest signal. Members get everything behind it.
- 2 dated, sourced signals
- 5 pieces of hard evidence
- 3 builders and where each falls short
- 2 charts and tables
- 8 linked sources
Already a member? Log in
Signals are specific dated events, each linked to its source and reviewed before entry. The read is editorial interpretation, not financial or investment advice. Do your own diligence before building or investing.