Matthew Widdop
Matty is a Digital SEO Executive at Intelligency, helping our clients to improve their digital performance. Matty is currently studying for his Level 3 Multi-Channel Marketer apprenticeship and has completed a Sport Journalism Degree at the University of Huddersfield.
What is Black Hat SEO?
Black Hat SEO is the practice of using undesirable techniques in order to boost the SEO performance of a website. This can include, but is not limited t,o keyword stuffing, hidden links, paid backlinks and more. Google’s algorithms have found a way to penalise sites using different Black Hat SEO tactics, while promoting sites that use White Hat SEO techniques such as creating quality content and focusing on user experience. However, there are certain individuals who will always be trying to trick search engines to artificially boost their rankings. In this article, we will look into new Black Hat SEO techniques that have emerged in 2025 and why you should steer clear of them.
How does Black Hat SEO work in 2025?
The main black hat SEO technique that has become prevalent in 2025 is AI poisoning. This is when users purposefully feed information to chatbots that is wrong or misleading in order to damage a brand’s reputation or offering. Up until now, it was believed that it would take a large amount of data to manipulate AI responses, as they have already been trained to their current understanding on vast amounts of data. However, a recent study from Anthropic shows that only a small amount (as few as 250) malicious documents can cause this Black Hat SEO technique to take effect on the user’s desired Large Language Model.
Here’s what the study had to say,
“Poisoning attacks can compromise the safety of large language models (LLMs) by injecting malicious documents into their training data. Existing work has studied pretraining poisoning, assuming adversaries control a percentage of the training corpus. However, for large models, even small percentages translate to impractically large amounts of data. This work demonstrates for the first time that poisoning attacks instead require a near-constant number of documents regardless of dataset size. We conduct the largest pretraining poisoning experiments to date, pretraining models from 600M to 13B parameters on chinchilla-optimal datasets (6B to 260B tokens). We find that 250 poisoned documents similarly compromise models across all model and dataset sizes, despite the largest models training on more than 20 times more clean data. We also run smaller-scale experiments to ablate factors that could influence attack success, including broader ratios of poisoned to clean data and non-random distributions of poisoned samples. Finally, we demonstrate the same dynamics for poisoning during fine-tuning. Altogether, our results suggest that injecting backdoors through data poisoning may be easier for large models than previously believed, as the number of poisons required does not scale up with model size, highlighting the need for more research on defences to mitigate this risk in future models.”
What this means for SEOs
Obviously, it goes without saying that manipulating large language models to try and gain some sort of edge on your competitors is not an encouraging technique, as not only is it self-serving, but it can also come back to be a negative for you as well. Firstly, if people start manipulating large language models on a large scale purely for selfish reasons, then these models will become largely unusable and have to be retrained, which is bad for everyone in the industry ultimately. Secondly, if Google starts penalising people who have been manipulating large language models and feeding them ill information, your site could be heavily penalised or, worse, altogether restricted from appearing on Google. As ever, making sure you follow Google’s guidelines and best practices when enhancing your site and its seo performance, while creating quality content, is the best and most surefire way to beat out your competitors in 2025.
