DenmarkTechnology and telecom2024 – ongoingTrustpilot Group
Machine learning fake-review removal and generative AI content moderation
Trustpilot said it removes more than 90% of fake reviews automatically with machine learning
Problem
The review platform has to keep fake and abusive content off the site across millions of reviews.
Approach
Trustpilot uses several machine learning techniques that look for signals such as unnatural language and review spikes, and the models are updated as manipulation tactics change. Since 2024 it has also used generative AI to flag likely breaches of policy on impersonation, abuse and misinformation. Human moderators review the flagged content and handle complex cases.
Results
Takeaway
Automated screening handles most fake-review removal, and people stay in the loop for flagged policy cases. Trustpilot gave no figure for the generative AI part.
Categories
- Industry
- Technology and telecom
- Business function
- Production and qualityFinance, risk and legal
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and MLGenerative AI
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
Trustpilot's own page on how it uses AI, reflecting practices as of June 2025. It says more than 90% of fake reviews are removed automatically by machine learning and that generative AI has flagged policy violations for human review since 2024.
Full details are in the original sources.
