Reading list
The sources behind the journal, numbered in the order I logged them. Journal entries link here.
Chen, M., Wang, X., Chen, K., & Koudas, N. (2025). Generative engine optimization: How to dominate AI search. arXiv. doi.org/10.48550/arXiv.2509.08919
GEO is the perfect example of how marketers influence purchasing decisions at their core. A classmate suggested I could run its method at a smaller scale instead of inventing a new one.
Zhang, D., Li, W., Niu, B., & Wu, C. (2023). A deep learning approach for detecting fake reviewers: Exploiting reviewing behavior and textual information. Decision Support Systems, 166, Article 113911. doi.org/10.1016/j.dss.2022.113911
Fake reviewers measurably hurt trust in a platform. That backs my choice of reviews as the 1 variable to isolate.
Capital One Shopping Research. (2026, March 3). Fake review statistics (2026): On Amazon & other websites. capitaloneshopping.com/research/fake-review-statistics
This gives me my headline numbers: roughly 30% of online reviews are fake, at an estimated $770.4 billion in unwanted purchases in 2025.
Luca, M., & Zervas, G. (2016). Fake it till you make it: Reputation, competition, and Yelp review fraud. Management Science, 62(12), 3412-3427. doi.org/10.1287/mnsc.2015.2304
Yelp filtered about 16% of restaurant reviews as suspicious, mostly at struggling businesses or ones with many competitors. Fraud is strategic.
He, S., Hollenbeck, B., & Proserpio, D. (2022). The market for fake reviews. Marketing Science, 41(5), 896-921.
He et al. documented an actual marketplace for fake Amazon reviews run through Facebook groups, with short-lived bumps in ratings and sales.
Federal Trade Commission. (2025, December 22). FTC warns 10 companies about possible violations of the agency's new Consumer Review Rule [Press release]. ftc.gov
Proof the regulators have already responded. The FTC now bans fake, insider, and AI-generated reviews, with penalties up to $53,088 per violation.
Yang, F., Chen, H., Yu, X., Zhang, M., Xiao, X., & Deng, M. (2027). A survey on fake review detection: From pre-trained language models to large language models. Information Fusion, 138, Article 104715. doi.org/10.1016/j.inffus.2026.104715
My map of what already exists: nearly 2 decades of fake-review detection, from classic machine learning to BERT to LLM detectors.
Mozilla. (2025, May 22). Investing in what moves the internet forward. The Mozilla Blog. blog.mozilla.org
Fakespot was the closest tool to my idea. Mozilla shut it down in 2025, probably because scraping at that scale cost too much to keep running.
Ott, M., Choi, Y., Cardie, C., & Hancock, J. T. (2011). Finding deceptive opinion spam by any stretch of the imagination. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (pp. 309-319). aclanthology.org/P11-1032
A classmate pointed me to this labeled deceptive-review corpus. It would let me test the tool on reviews already marked real or fake.
Open-access article shared by my mentor on Sep 16, 2026. doaj.org
Reading for fake review detection. Title and authors go here once I have read it.
Information Fusion article shared by my mentor on Sep 16, 2026, via the ASU library. sciencedirect.com
The second reading for the same task. Likely the same survey I cite as Yang et al. above. If so, the 2 entries merge.
Little, K. (2026, September 23). Free trials are costing Americans $540 a year, on average, survey finds. Yahoo Finance. finance.yahoo.com
A survey of 2,000 consumers puts forgotten free trials at about $540 a year. Not about ratings, but the same family of tactics, so it helps me decide where my problem statement stops. It is a news summary of a commercial survey, so a pointer, not evidence. The 4 entries after it are the stronger sources on the same topic.
Luguri, J., & Strahilevitz, L. J. (2021). Shining a light on dark patterns. Journal of Legal Analysis, 13(1), 43-109. doi.org/10.1093/jla/laaa006
To read: 2 survey experiments that offered a free trial with monthly billing afterward, with and without manipulative design. The authors report that hiding the terms about doubled sign-ups, while a countdown timer had no clear effect.
Mathur, A., Acar, G., Friedman, M. J., Lucherini, E., Mayer, J., Chetty, M., & Narayanan, A. (2019). Dark patterns at scale: Findings from a crawl of 11K shopping websites. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), Article 81, 1-32. doi.org/10.1145/3359183
On my list: a crawl of about 11,000 shopping sites that puts hidden subscriptions and hard-to-cancel flows in the same taxonomy as countdown timers and low-stock messages. It shows where free trials sit among the tactics I study.
Federal Trade Commission. (2026, March 11). FTC seeks public comment in response to advance notice of proposed rulemaking regarding negative option marketing practices [Press release]. ftc.gov
To read: the FTC asking the public how to regulate free trials and auto-renewals, after a court vacated its click-to-cancel rule in July 2025. When I checked on Sep 28, 2026, no federal click-to-cancel rule was in force.
New York City Department of Consumer and Worker Protection. (2026, October 1). NYC's "Click to Cancel" rule is now in effect. nyc.gov
Saved from a post on X. The first city rule that makes cancelling a subscription as easy as signing up, in force since Oct 1, 2026, with penalties from $525 per violation. It sits next to the FTC entry above: no federal click-to-cancel rule, but a city one.
The Wall Street Journal. (2026, October). Restaurants are using AI to advertise their food and it's making people sick. wsj.com
To read: restaurants and ghost kitchens putting AI-made food pictures on delivery apps, and diners noticing the dish does not match. A tactic that is not on my list yet: a picture of the product that no camera took. Paywalled, so the author and the exact date are still to confirm.
Hou, Y., Li, J., He, Z., Yan, A., Chen, X., & McAuley, J. (2024). Bridging language and items for retrieval and recommendation (Amazon Reviews'23) [Data set and preprint]. McAuley Lab, UC San Diego. arXiv:2403.03952. amazon-reviews-2023.github.io
Suggested by my research class professor. 571.54 million Amazon reviews from May 1996 to September 2023 across 33 categories, with a verified-purchase flag and timestamps. No fake labels and no license stated on the page. If the category stays a day-to-day product, the baseline comes from here. I have not pulled the toothpaste rows yet.
Yelp. (n.d.). Yelp Open Dataset [Data set]. Retrieved October 6, 2026. business.yelp.com
Also suggested by the professor. 6,990,280 reviews of 150,346 businesses in 11 metro areas, for educational use. Restaurants and local businesses, not products, so for me a place to test methods rather than my data.
He, S., Hollenbeck, B., & Proserpio, D. (2022). Fake reviews data: Amazon reviews labeled fake or real [Data set]. GitHub. github.com/bretthollenbeck/fake-reviews-data
The data behind He et al. (2022) above, MIT license: reviews and reviewers labeled fake or real for products whose sellers were caught recruiting reviews on Facebook, plus which reviews Amazon deleted between 2020 and 2023. The only market-sourced Amazon ground truth I found in the open. Not counted yet, and the share of day-to-day products in it is unknown.
Salminen, J., Kandpal, C., Kamel, A. M., Jung, S.-g., & Jansen, B. J. (2022). Creating and detecting fake reviews of online products. Journal of Retailing and Consumer Services, 64, 102771. doi.org/10.1016/j.jretconser.2021.102771
To read: 40,000 Amazon reviews, half written by GPT-2 and half real, across 10 categories, with the data on OSF (osf.io/tyue9). The label means machine-written, not bought, so it covers the AI text case only.
Zhang, Y., Hao, S., & Wang, H. (2020). Review trade: Everything is free in incentivized review groups. In Security and Privacy in Communication Networks (SecureComm 2020). Springer. Data: github.com/zhangyubao5/incentivized_review_group
To read: 39,880 Amazon reviews traced to Facebook, Douban and WeChat groups that trade free products for reviews, 451 products, 76% of them 5 stars. Positives only, and no license on the file. Incentivized is not the same as fake, which matters under the FTC rule below.
Federal Trade Commission. (2024, August 14). Federal Trade Commission announces final rule banning fake reviews and testimonials [Press release]. ftc.gov
The rule behind the 2025 warning letters above (16 CFR Part 465): fake or false reviews and testimonials, buying positive or negative reviews, undisclosed insider reviews, company-controlled review sites, review suppression and fake social media indicators. The professor's search terms pointed here.