Singapore's CCCS acted against Reputifly and 45 businesses over AI-generated fake reviews built to evade detection
Singapore's Competition and Consumer Commission (CCCS) took action against Reputifly Pte. Ltd., which sold AI-generated fake reviews trained on real Google reviews to evade platform detectors. Human proxies recruited on Telegram posted the reviews from their own accounts on Google, Facebook, Tripadvisor, Carousell, Yelp and Trustpilot. CCCS identified about 100 businesses that had obtained fake reviews from Reputifly. Of 47 examined in the first phase, 45 admitted buying them and must post public apologies for six months, while the rest remain under investigation. Reputifly admitted aiding and abetting unfair trade practices, agreed to stop, and must donate proceeds to charity.
What the AI did
A Singapore company, Reputifly, used an AI system it built to write fake customer reviews, using real Google reviews as training material so the fake ones would get past the platforms' automated detection. People recruited through Telegram channels and paid by Reputifly then posted these AI-written reviews from their own accounts on Google, Facebook, Tripadvisor, Carousell, Yelp and Trustpilot.
First reported October 1, 2026 · Added to the register October 8, 2026 · 1 source
- Developer
- Reputifly Pte. Ltd.
- Country
- Singapore
- When it happened
- September 30, 2026
- First reported
- October 1, 2026
What this means for you
Could this affect you?
AI-written fake reviews were posted on six major review platforms for businesses ranging from medical clinics and law firms to restaurants, hair salons and funeral services, sectors where consumers turn to star ratings to judge unfamiliar providers. Forty-five businesses that bought them must post public apologies for six months, and other buyers remain under investigation.
What to check
- Prohibit buying or soliciting reviews from third-party providers.
- Audit marketing vendors for fake-review services.
- Consider coming forward to CCCS early if your business has used such services, as cooperation is viewed favourably.
Areas of your AI programme this touches
Timeline
- September 30, 2026Happened
- October 1, 2026First reported
Every fact and its source (10)
- Businesses began posting apologies2026-09-30“Singapore's competition watchdog forced 45 businesses to begin posting public apologies on September 30, 2026”[1]
- Businesses identified as buyersAbout 100“CCCS identified approximately 100 businesses that had obtained fake reviews from Reputifly”[1]
- Businesses that admitted buying fake reviews45 of 47 examined in Phase 1“In Phase 1, 47 businesses were examined; 45 admitted to purchasing fabricated reviews and gave formal undertakings to the commission.”[1]
- Example package price35 five-star reviews for S$219“One of those packages sold 35 five-star reviews over seven days for S$219 (approximately $172 USD)”[1]
- Provider admissionReputifly admitted aiding and abetting unfair trade practices“Reputifly itself admitted to aiding and abetting unfair trade practices and agreed to stop providing fake-review services.”[1]
- Remedy for businessesSix-month public apology“publish a public apology for a continuous six-month period”[1]
- Remedy for providerDonate proceeds to charity, no tax deduction“the company must donate proceeds to a registered charity and is prohibited from claiming tax deductions on that donation”[1]
- Platforms where fake reviews were postedGoogle, Facebook, Tripadvisor, Carousell, Yelp, Trustpilot“Reputifly recruited human proxies through Telegram channels, paying them to post AI-crafted reviews from their own accounts across Google, Facebook, Tripadvisor, Carousell, Yelp, and Trustpilot.”[1]
- Operator of ReputiflyJulian Tung Yan Kai“their supplier, Reputifly Pte. Ltd., operated by Julian Tung Yan Kai”[1]
- AI training methodTrained on real Google reviews to evade AI classifiers“it used real Google reviews as training material to calibrate the output”[1]
Sources
- AI Fake-Review Provider Trained on Real Google Ratings Hit by Singapore Record Enforcementtechtimes.com · October 1, 2026
How this record is classified. Severity M (2/5): harm to individuals, or a policy breach with limited reach. Evidence: Confirmed, meaning a company, official or court statement.
The facts in At a glance and in Every fact and its source are quoted from the sources listed. The summary, What the AI did and What this means for you are written from those sources and checked against them automatically before publication. Records reflect what has been disclosed, not everything that has happened. Incidents are found through news feeds and the GDELT Project.
