Experts express concern over the potential impact of AI on terrorism through autonomous car bombs and online recruitment.

There are concerns among experts that AI could be exploited by terrorists for innovative and challenging purposes, such as enhancing their explosive delivery techniques and intensifying their online recruitment efforts.

AI can pose a grave threat if used with malicious intent, as stated by Antonia Marie De Meo, director of the United Nations Interregional Crime and Justice Research Institute, in a report examining how terrorists might exploit AI.

"This tool, with a history of success in cybercrime, could potentially aid in terrorism and violent extremism, including self-driving car bombs, enhancing cyberattacks, and spreading hate speech or inciting violence online," she stated.

To stay ahead of terrorists utilizing AI, law enforcement must remain at the forefront of AI technology, according to the report "Algorithms and Terrorism: The Malicious Use of Artificial Intelligence for Terrorist Purposes."

De Meo wrote that it is our aspiration for this report to initiate a discussion on the harmful use of AI for terrorist activities.

The task of staying ahead of terrorists and anticipating their use of AI is a challenging one, as it requires not only coming up with innovative ways to use AI, but also finding ways to prevent others from using those same methods.

The study, "Emerging Technologies and Terrorism: An American Perspective," which was conducted in collaboration between NATO COE-DAT and the U.S. Army War College Strategic Studies Institute, supports the claim that terrorist groups are using these tools for recruitment and attacks.

In the era of fast technological advancement, the boundary between fact and fiction becomes indistinct, prompting governments, businesses, and academia to collaborate in developing ethical guidelines and rules.

The authors emphasized the importance of national responsibility in combating terrorism and advocating for collective strength against the looming specter of technology-driven threats as geopolitical tides shift.

OpenAI's ChatGPT can be used to improve phishing emails, plant malware in open-coding libraries, spread disinformation, and create online propaganda.

The authors noted that cybercriminals and terrorists have become skilled in utilizing platforms and large language models to create deepfakes or chatbots hosted on the dark web to obtain sensitive personal and financial information or to plan terror attacks or recruit followers.

"As AI models become more advanced, the likelihood of malicious use is likely to increase, and it will be crucial to ensure transparency and controls over how sensitive conversations and Internet searches are stored and distributed on these platforms or through large language models."

This year, West Point's Combating Terrorism Center released research on improving planning capabilities for terrorist attacks, rather than just enhancing their current strategies.

The authors examined the potential consequences of inputting commands that enable these systems to bypass their standards and policies, thereby allowing the base model to deliver extremist, illegal, or unethical content.

"The authors examined the potential use of five large language models by extremists through the use of multiple accounts, specifically in the areas of training, operational planning, and propaganda development."

The study showed that Bard was the most resistant to jailbreaking, with ChatGPT models following closely behind. In more than half of the cases, indirect prompts were found to be sufficient to bypass the guardrails.

The research concluded that jailbreak guardrails require continuous evaluation and "enhanced collaboration between private and public sectors," including academia, tech companies, and the security community.