Google has posted a new policy saying AI-generated reviews are against its policies, disallowed and considered spam. If you find such content, Google said you must mark it as spam in your feed with the is_spam attribute.
Google posted the update over here, it reads, "Automated Content: We don’t allow reviews that are primarily generated by an automated program or artificial intelligence application. If you have identified such content, it should be marked as spam in your feed using the is_spam attribute."
How does one identify if user-generated reviews are generated by AI is another story. But if you are posting your own reviews using AI, you must mark them as spam if you are using Google Merchant Center.
Now, Google Search has been very lax about content generated with AI in general. Google is okay with content being generated by AI as long as it is useful and helpful. This policy somewhat goes against that, although, these are reviews and I guess Google expects humans to leave reviews, not robots.
Google also clarified its other policies, here are those clarifications:
- Spam: We don’t allow spam content. Ensure any content known to contain irrelevant, repetitive, or nonsensical text is marked as spam in your feed using the
is_spamattribute. - Dangerous products or acts: Don’t submit reviews of regulated products that can cause damage, harm, or injury. For example, don’t submit reviews of guns, tobacco products, or regulated drugs. Additionally, don’t include content that depicts or provides instructions to complete activities that are dangerous and/or widely illegal.
- Phone numbers, email addresses or URLs: Don't submit phone numbers, email addresses and links to other websites in the review content.
- Personal and confidential information: Don't submit reviews that contain personal information, including but not limited to full names, credit card information, national insurance number, driving license information, etc.
- Keep it clean: Don’t submit reviews that use obscene, profane, or offensive language or include content that depicts scenes of violence, serious injury or death, contains personal attacks, or are defamatory.
- Conflict of interest: Don’t submit reviews that were paid for or are otherwise inauthentic. Reviews are most valuable when they are honest and unbiased. We remove reviews that we believe have conflicts of interest and/or have been written by employees or people with a vested interest in the product. Only submit reviews that were honestly solicited from customers who made a purchase.
- Illegal content: Don't submit reviews that contain or link to unlawful content, such as links that facilitate the sale of prescription drugs without a prescription, illegal drugs, counterfeit products, or illicit weapons. This content is not allowed.
- Malware & Phishing: Don’t submit reviews containing links to malware, viruses, or other harmful software. We don't allow phishing or other content that harms or interferes with the operation of the networks, servers, or other infrastructure of Google or others.
- Copyrighted content: Don't submit reviews that infringe on others' rights, including copyright. For more information or to file a Digital Millennium Copyright Act (DMCA) request, review our copyright procedures.
- Trademark infringement: Don’t submit content that uses a trademark in a way that’s likely to cause confusion about the origin of the product being reviewed.
- Plagiarism: Don’t submit reviews containing appropriation of content created by another person or entity.
- Sexually explicit material: Don't submit reviews that contain sexually explicit material. We also don't allow reviews that sexually exploit children or present them in a sexual manner. For this type of content, we remove the review, shut down the product reviews feed, and send a report to the National Center for Missing & Exploited Children (NCMEC) and to law enforcement. This content is not allowed.
- Hate speech: Don’t submit reviews that advocate against groups of people based on their race or ethnic origin, nationality, religion, disability, gender, age, veteran status, sexual orientation, or gender identity. Violent language, terrorist content, or content intended to incite and encourage violent acts or extremist behavior, are not allowed in reviews.
- Cross promotion of other products/websites: We don’t allow reviews that include promotional content for other unrelated websites, products, or services. If you have identified such content, it should be marked as spam in your feed using the
- Off-topic reviews: We may remove reviews that primarily seek to discuss other topics unrelated to the product itself. This includes comments about shipping and delivery, experiences with particular retailers, political or social controversy, attacks on others, or don't represent a first-hand experience with the product. This also includes merchant responses to reviewer comments.
- Impersonation: We don't allow reviews from people claiming to be someone that they aren't.
- Language: Reviews should be submitted in the original language they are written in. Google will display reviews in the original language with an option on the Shopping page to be translated to the user’s language.
- Duplicate content: Each review should only appear in one feed and only once in that feed. If a review applies to multiple variations of a product, this should be designated by including multiple unique product identifiers and URLs with the review. The same review shouldn't appear in feeds from different retailers. The retailer should own the review and shouldn't send Google content syndicated from other sources. In the case of duplicate reviews in multiple feeds, we may blocklist one or more of the reviews or entire feeds.
Finally, Google posted guidance on how it can enforce these policies:
- We use a combination of automated and human evaluation to ensure that content and reviews comply with our policies. Our enforcement technologies use machine-learning algorithms to help protect our merchants and users by keeping our shopping platforms safe. More complex, nuanced, or severe cases are reviewed and evaluated by our specially trained experts who conduct content evaluations that might be difficult for algorithms to perform alone, for example, because an understanding of the context of the piece of content is required.
- We take action on content and reviews that violate our policies. This may include disapproving violating content or a violating review, as well as issuing warnings or suspending accounts for repeated or egregious violations. We take repeat violations of our policies seriously.
- When an image is flagged for a policy violation, we will now also block the associated review content.
Forum discussion at Twitter.
