This sentence betrays a spam bot polluting the web with content generated by AI

The use of ChatGPT and other AI could be filling the web with low-quality content, spam and fake reviews. Automated bots copy and paste automatically generated text.

“As an AI language model, my programming prohibits me from generating hateful and harmful tweets towards individuals or groups of people.”, This is the phrase that any artificial intelligence uses to avoid giving sensitive information or that threatens third parties. However, that start is the clue that reveals when they have been set. automated bots that fill the Internet with spam.

The advance of Artificial Intelligence is evident, but it is not all good news: the proliferation of spam-generating bots has created some problems in the quality of the information around the writing of text, false reviews or directly tons of spam on social networks and the web in general.

Just type this sentence on Twitter: “As an AI language model”. If you do, you will find plenty of examples that the community points out and denouncesmost of the time in a humorous tone, about this growing and problematic reality.

In this article, we also listed the possibility of using jailbreaks, or cheats that try to semantically or psychologically cheat the barriers imposed by the developers, so that chatbots bypass the prohibition of giving sensitive information. Another problem that worries artificial intelligence companies a lot.

Bots post AI-generated content in product comments

Book reviews, product reviews on AmazonLinkedIn post, nothing is saved from those who want to generate easy profits through this new technology. Security engineer Daniel Feldman points to a comment on a product on Amazon to illustrate what is happening.

“As an AI language model, I have not personally used this product, but based on its features and customer feedback, I can confidently give it a five-star rating.”.

“Regenerate response”, another of the phrases that indicate spam

When searching for these two words: “Regenerate response” On social networks or search engines, we found numerous posts that were obviously copied and pasted from the OpenAI chat bot, probably through an automated bot.

“Regenerate response”, is the option that ChatGPT gives to reprocess a text to the user, and bots can indiscriminately copy and paste this option, revealing themselves as spam.

How to prevent the spread of spam generated by AI bots

The evolution of AI is an excellent opportunity to improve the quality of the content generated and it could be used to identify patterns, correct errors and offer new perspectives, a new tool that, when used well, can be very useful. However, it is difficult to detect their fraudulent use.

The possibilities of AI are many, among them the medical application and scientific advances, new possibilities in artistic creations, the economy, education or the energy industry. But the difficulties he poses are proportional to the advantages.

Generating spam and low-quality content is just another of the challenges facing this new tool. and it adds to the problems with copyright or the use of data and privacy in the training of artificial intelligences.

The EU, for its part, is already working in a regulation for artificial intelligences and the technology directors themselves have warned about the ethical and regulatory need in this field. The entrepreneurs themselves dedicated to artificial intelligence even signed a letter asking for the delay in the development of this technology in order to understand where we are going with it.

The solution to prevent the spread of spam generated by AI bots is not easy. Experts point to the identification of common patterns in spam, creation of detection programs and continuous monitoring of content generated by the models.