Including tips and recommendations to help website owners navigate the AI-UGC triple (detection, volume and speed of low-quality AI user-generated content).)
The term “low-quality user-generated AI content” is a chatty term, but it is an important issue for website owners who allow UGC to be published on their websites. And I’ve encountered several difficult situations recently after intensively analyzing September’s helpful content update (HCU). I don’t think anyone has covered the topic of the impact of UGC with the helpful content update, so I wanted to cover it briefly in this post.
Since the first HCU launched in August 2022, I have seen many examples of low-quality AI content hit hard. I’ve mentioned this several times when reporting on drops with the HCU in August 2022, the HCU in December 2022, and now the more aggressive HCU(X) in September. However, this was more about the site owner using AI extensively to publish low-quality (and unhelpful) content on their own site. Therefore, over time, the websites became loaded with a lot of lower quality AI content and were attacked by the HCU.
What I’m referring to today is roughly other people Adding lower quality AI content to your website via UGC. In other words, what if people submitting user-generated content used AI to quickly create that content? And what if that content is lower quality, not edited to add value, not refined to be truly insightful, etc.? Yes, this can be problematic, and website owners who allow UGC are starting to notice this problem increasing. And some sites that contacted me after HCU in September were hit very hard and lost a lot of search visibility.
A reminder of Google’s stance on AI content: It’s about quality.
It’s important to remember this Google is not against all AI content. It is against it low-quality AI content. So if someone uses AI to help create content and edits and refines that content to make sure it’s accurate, valuable, etc. then that could be totally fine. But if someone simply exports AI-generated content and publishes it at scale, then they are setting their site up to be wiped out by either the helpful content update or even a major core update.
Here is a paragraph from Google’s post on AI content where Danny Sullivan explains this AI content with the main purpose of manipulating the ranking in the search results is what they have a problem with:
The Insidious Development of Low-Quality User-Generated AI Content:
In my opinion, and based on what I’ve seen while analyzing many websites affected by HCU, low-quality AI content can be incredibly dangerous at scale. And if UGC is a core part of your site’s content, you need to keep a close eye on quality. This has been the case with UGC for a very long time, but now it’s much easier for users to use AI authoring tools to write answers in forums, Q&A sites, and more.
And just because someone else posted the content on your site, you are still responsible. Google has explained this many times over the years. Below I’ve included just one of several tweets I’ve shared via Google over the years explaining that UGC is taken into account when assessing quality. If it is on your site and indexed, it will be included in the quality score.
Moderate UGC strong:
Since the medieval Panda days (circa 2011), I have always stated that website owners need to heavily moderate user-generated content. If not, lower quality content can creep in and build up over time. And when that happens, prepare your site to be hit hard by a big algorithm update. So my advice now is the same, but with a slight AI twist. I would now also look for lower quality products AI content and not just your typical UGC spam. The leading AI content discovery tools are constantly improving and can help flag content with a high probability of being created using AI.
Below I provide some tips for website owners who may be dealing with lower quality AI user-generated content. Again, I’ve been contacted by several websites that have had this issue in the last year or so…
Detection, Volume and Pace: The AI UGC Triple
In addition to identifying AI-generated content, you can also check the number of responses from users and the speed at which those responses are delivered. A site owner recently reached out to me and saw a user who sent a lot of replies in a short period of time. When these answers were run through an AI content discovery tool, they all came back with a 100 percent probability that the content was created by AI. For example, you can see two of these submissions below.
The combination of volume, pace, and fairly obvious AI content produced a good example of a user using AI to post a lot of UGC on the website in question.
Note that this is a large site, so this one user would likely have little impact. However, if other users did the same thing over time, it could well cause problems from a quality perspective. The website owner was smart to contact me about this. They took care of this user and potentially developed a pattern identification process that could help uncover people trying to game the system with lower quality user-generated AI content.
And remember, HCU in September seemed to bring UX into the equation (which I clearly noticed when analyzing many affected websites). Therefore, user experience barriers such as aggressive ads, pop-ups, interstitials, and more can contribute to “unhelpful content.” So if you combine low-quality user-generated AI content with terrible UX, you could have a serious problem with the HCU traveling the internet. Caution.
Conclusion: A warning for website owners who accept UGC.
Again, I want to cover this topic for any website that publishes user-generated content. Even if you have a larger website with a ton of content, I would be very cautious about allowing low-quality content of any kind on the site. And now that AI content generation tools are so easy, you should be careful about allowing low-quality AI user-generated content on the site. Again, several companies have come forward after being hit hard, and that was clearly part of the problem.
I’ll end this post with some tips and recommendations for website owners:
- Hard moderate and with an AI twist: If you accept user-generated content on your site, you should already be heavily moderated in terms of quality. Now you should also look out for lower quality products AI Content that is submitted.
- Speed of submission: A red flag could be the pace of submission by certain users. For example, I explained earlier that a company reached out to me after reporting submissions that seemed a bit unusual… When we were dealing with this user, there were a number of submissions in a very short period of time. And after running these submissions through AI content discovery tools, there was a 100% chance that these submissions were created via AI.
- Test laboratory: I recommend setting up a testing lab that leverages multiple AI content recognition tools. The tools are not perfect, but they can detect well inferior quality AI generated content. Establish a process to review flagged content. Next, I’ll discuss APIs that can help with bulk content verification.
- APIs are your friend: Some AI content discovery tools have APIs that allow you to inspect content in bulk. If you have a large website with a lot of user-generated content, using an API could be a smart solution. For example, GPTZero, Originality.aiand others have APIs you can use to inspect content at scale.
- AI Content Policy for User Generated Content: Set a policy for AI content that users can access and easily understand. Make sure users understand what is allowed and what is permitted not allowed, including what you will do if low-quality AI content is detected.
- When it comes to indexing, be quick, be decisive: And finally, deal with AI submissions quickly and ensure they don’t get indexed (or stay unindexed). As I’ve explained many times in posts and presentations about important algorithm updates, Google’s quality assessment takes all indexed pages into account. So focus on what I call “quality indexing” and ensure that lower quality user-generated AI content is not indexed.
Summary: Pay attention to UGC with an AI twist.
For websites that accept user-generated content, it has always been important to moderate that content heavily from a quality perspective. And with AI tools for content generation, dealing with UGC has now become more difficult and moderation has become a little more difficult. I recommend reading the tips and recommendations I’ve provided in this post to create a process for flagging potential issues and then quickly resolving those issues. This is the best way to maintain a strong level of “quality indexing,” which is important to avoid issues based on major algorithm updates such as the helpful content update and comprehensive core updates.
GG