
AI chatbots are now one of the most popular solutions to help organizations sail through these uncertain times. Businesses are always looking for ways to shave off operational costs as well as optimize them. They will choose to implement live chatbot software for the company to achieve both of these objectives. Recently, chat GPT made quite an impression upon its launch and soon became a sensation among users. Capable of offering human-like conversations with new generations of conversational AI models, it immediately caught public attention.
Turing test results for chat GPT are promising as a true AI, showing the solution has some serious capabilities. It cannot only fool humans into thinking they are conversing with a real person, but it also shows the ability to write convincing reports, articles, essays, and even novels. For example, it created an introductory paragraph of an article for a reputable research firm. Even if writers can detect it, the AI solution’s effort was indeed commendable.
ChatGPT can be used for various intelligent activities like personalizing education, writing newspaper articles, and even writing computer codes. Though the code it generates may have vulnerabilities or security issues, texts might have factual errors or be unable to convey the subject matter properly. Still, it can learn over time from all the information available on the internet and have incredible potential to alter the entire business and service landscape.
Apart from professionals from consulting firms, they may not recognize an AI application that created it. Immediately getting attention from all corners of the world, it reached one million users, but it also raised the alarm for risks and security reasons.
Indeed, there are plenty of opportunities to utilize it for good purposes as well as exploit it with malicious intentions.
ChatGPT and lurking cybersecurity risks
Developing a live chatbot software for a company with capabilities like lyrics for songs to academic essays can revolutionize your business. However, chat GPT can write codes like humans, and the cybersecurity community is already skeptical about it. According to reports, Google has issued a code red to retain their search business as this AI is believed to be able to reinvent or even replace traditional internet search.
An Israeli cybersecurity firm has demonstrated how Chat GPT can be used for creating phishing emails. It was created by utilizing OpenAI’s code-writing system, Codex, carrying a malicious payload. According to some experts, it has the potential to alter the cyber threat landscape significantly, and it is another step toward the dangerous evolution of effective and sophisticated cyber capabilities. Unfortunately, other experts were also able to create phishing emails. Though first, it denied the request, slightly altering the request let them bypass the built-in safeguard measures.
Some believe it can be used by cybercriminals who are not native English speakers to create legitimate-sounding phishing emails conveniently. It can again become one of the most effective ways to launch a ransomware attack. It can create realistic and interactive content for compromising the business email. More concerning is that it can be used to launch attacks over other channels like Facebook Messenger, WhatsApp, and other chat applications.
So, how serious is it?
The hackers can utilize the code-writing capability of this intelligent application with poor or even no coding experience for malicious purposes like spying, stealing, ransomware attack, and more. For example, a thread published in an underground hacking forum revealed that the publisher was trying to recreate the techniques described in a research publication for creating a common malware with ChatGPT. As a result, he successfully created a Python-based data stealer capable of searching common file types, copying them into a random folder inside the Temp folder, ZIP them, and uploading them over the FTP server.
He also showed a simple Java snippet created using AI that can be modified to run any program, including the common malware families. What is more alarming is another cybercriminal with a handle called USDoD submitted a Python script that he apparently wrote. A peer recognizing the style of the code similar to openAI probed, and the threat actor acknowledged that ChatGPT gave him a good helping hand in finishing the script. The script is capable of encrypting a user’s machine, and it can be turned into ransomware by rectifying the code a little.
Also, many cybercriminals are trying to use their natural responses to do fraudulent activities. A Finnish security company has warned that cybercriminals will use this technology. The generation of versatile natural language text from little inputs will inevitably attract these actors with malicious intents to exploit this capability. They will use it to spread fake news, scams, or misinformation with credible, compelling text created with superhuman speed and little effort. These threat actors are trying to produce random art for a chapter for an e-book and sell them online via reputable marketplaces.
Final thought
Implementing live chatbot software for the company is normal for organizations to utilize the advantages of AI solutions. However, ChatGPT bared the vulnerabilities of these AI-enabled solutions. With the large volume of information, available online smart AI chatbots can perform many functions better than their human counterparts. It is the way to revolutionize the workplaces by automating manual tasks, and limitless potential is getting uncovered with these new breeds of solutions accumulating other advanced technologies. Whether it is summarizing reports or writing codes, it can draw from existing materials and learn. But ChatGPT revealed a major weakness of these AIS, and industries must learn how to safeguard these solutions from potential attacks faster to ensure full utilization of this revolutionary software technology.
