- This attack, unlike prompt injection, focuses on the AI nature.
AI Squatting: Artificial Intelligence has developed very rapidly in today’s time. Today people are using AI even for small tasks. AI is being used today from big private offices to government offices also. But along with this, AI is also being used in cyber attacks. Now cyber thugs are working to cheat people through AI. In this the name AI Squatting has emerged. Let us know what is the whole matter.
What is AI Squatting?
According to the information, Artificial Intelligence based chatbots and large language models (LLMs) sometimes give the name of such websites or software packages which do not exist in reality. This mistake is called AI Hallucination.
Let us tell you that earlier it was considered to be only a technological shortcoming of AI, but now cyber criminals are taking advantage of these mistakes and creating fake websites and fake software. If a user trusts AI’s advice and reaches those links or downloads the package, his data may be stolen or malware may be installed on the device.
Threat to companies
According to the information, today big companies are rapidly using AI tools for writing code, installing software, searching documents and IT automation. If AI recommends any fake website or fake software package and is used without checking it, then the entire network of the company can become vulnerable to cyber attack.
How AI Hallucination is becoming a threat
According to cyber safety experts, attackers are now pre-registering the names of those websites and packages which AI often suggests by mistake. Later, when a user asks the same question to the AI, the AI advises the same fake website or package and the user directly gets trapped in the trap of cyber criminals.
What is Prompt Injection?
Now let us tell you that earlier the most discussed issue in AI security was Prompt Injection, in which AI’s reaction was changed by giving wrong instructions. But AI Squatting is a completely different technology. There is no need for any prompt in this. Also, AI is not hacked or jailbroken. Apart from this, no technical weakness of AI is taken advantage of.
Let us tell you that now cyber attacks are not limited only to software weaknesses or user mistakes. AI Squatting shows that attackers are now targeting the nature of AI itself. Therefore, blindly trusting the website, download link, software package or repository that AI tells you can prove to be dangerous.
Also read:
What happens to the smartphone battery after it gets damaged, does it become E-Waste or is it recycled?
