Scammers: They use artificial intelligence to fuel their scams (ChatGPT)
Artificial intelligence (AI) has recently become an integral part of our society. The arrival of ChatGPT (designed by the American company OpenAI), a conversational agent that can generate written content in response to Internet users' queries, accessible to all, is now being used by students, in the office and in everyday life. It has become the essential tool of our time. Unfortunately, this AI is not only used for positive purposes, as many crooks and hackers take advantage of it to sublimate their scams and use it for harmful purposes.
Hackers get to grips with ChatGPT
According to a study carried out by CheckPoint, hackers have used ChatGPT to create malware and write content for phishing purposes. Hacker software has also been discovered that can be used to steal a host of documents in various formats (PDF, photos, etc.) from any Windows computer. The aim? To avoid being detected by anti-virus software.
They also use ChatGPT to write fake content or create fake sites. Because of ChatGPT's ability to write, scammers can use it to create beautiful phrases and speeches to feed their fake sites. These nice phrases will fool site visitors into thinking they are dealing with a genuine offer.
Phishing attempts
Hackers can also use ChatGPT in phishing campaigns.
These phishing attempts enable them to target a large number of people, taking advantage of data leaks that expose a multitude of emails in the darkweb. In the event of phishing, you should immediately check the e-mail address you have received to ensure that it is indeed that of your supplier.
Algorithms and fraudulent e-mails
CheckPoint has discovered that an algorithm can be created from a simple written request on ChatGPT. Scammers simply need to specify the function of the desired algorithm. These crooks use ChatGPT to create a Python script that can be used to block a user's data and demand a ransom in exchange for the key. They also create fraudulent e-mails that can be very misleading, as they are very well written and argued.
Here is an example of a fraudulent e-mail:
Identify texts created with AI
There are several tested and approved analysis tools for spotting texts that have been created using AI. The first is GPTZero Classic, which is easy to use. Simply copy and paste the paragraph or text you want to check, up to 5,000 characters long. Then press Ctrl+Enter. Thanks to this analysis tool, you can assess the probability of a text being generated by GPT technology.
Here's an overview of the tool that analysed a text.
The second tool is Draft & Goal. For the moment, it is limited to an analysis of English texts, but French content is being developed. This will help you avoid the scams created by ChatGPT.
Below is a preview of the Draft&Goal tool analysing a text.
Other elements can be used to identify a text created by an AI. Firstly, ChatGPT's knowledge base ends in 2021. If the text refers to events that took place in 2022 or 2023, it is quite possible that the text was written by a human. The texts generated by ChatGPT can look very similar. Grammatical constructions are identical when a text is written by ChatGPT, so it is easy to identify a text written by ChatGPT.
To find out more about scams, read our articles on scams on Facebook Marketplace, or on scams between individuals.
It is becoming increasingly important to secure your transactions between private individuals
Technology doesn't just bring scams, it also brings significant improvements in payment security. That's why we've created Obvy, the only payment method that can secure all your transactions between individuals, goods and services, from €1 to €200,000 and on all platforms.
Obvy also enables users' identities to be verified securely, guarantees payment by hand or by delivery, and offers secure services based on cutting-edge banking technologies.