DistantNews
Support us
🇦🇺 Australia /Technology

Terrorist use of AI 'major national security concern', report finds

From ABC Australia · () English

Summarized and contextualized by DistantNews.

At a glance

News Named sources Context piece
  • A report highlights the significant national security risk posed by the exploitation of artificial intelligence by violent extremists.
  • AI models, including ChatGPT, were found to provide information that could be used for terrorist acts, with safeguards often easily bypassed.
  • The report calls for urgent action to improve the security of AI models, particularly open-source versions, to prevent their misuse.

Artificial intelligence presents a growing threat to national security, as extremist groups may exploit its capabilities for malicious purposes, according to a new report. The study flags the potential for AI to be used in acts of terrorism, a concern underscored by a recent incident in Queensland where a teenager was charged for allegedly creating violent scenarios on an AI platform.

AI models posed a greater risk than a simple web search because they were designed to persuasively engage in back-and-forth conversation.

— Adam HadleyThe executive director of Tech Against Terrorism explained the unique danger posed by AI.

The report, released by the United Nations-supported Tech Against Terrorism, involved testing 27 AI models. Testers asked for information relevant to carrying out terrorist acts, including instructions for making explosives and 3D-printed firearms. ChatGPT, the most popular generative AI system, reportedly refused only 48 percent of over 2,000 requests, indicating a significant vulnerability.

Adam Hadley, executive director of Tech Against Terrorism, emphasized that AI models pose a greater risk than simple web searches because they are designed for persuasive, conversational engagement. He compared AI platforms to instructors or mentors, capable of providing advice on dangerous topics. "A model that will iterate with a user, answer follow-up questions and refine an approach is behaving less like a manual and more like a coach," Hadley stated.

A model that will iterate with a user, answer follow-up questions and refine an approach is behaving less like a manual and more like a coach.

— Adam HadleyHadley illustrated how AI can act as a dangerous mentor to users seeking harmful information.

The report found that the safety features of common AI models are easily breached, and urgent action is needed to enhance the security of open-source AI models. These models, available for download from various developers, can have their safeguards removed entirely. "We do not accept that developers have an unqualified right to distribute technology that can be used to kill people," Hadley asserted, urging governments to protect their citizens from recklessly released technology.

We do not accept that developers have an unqualified right to distribute technology that can be used to kill people.

— Adam HadleyHadley argued for government intervention to regulate the distribution of potentially dangerous AI technology.
DistantNews Editorial

Originally published by ABC Australia. Summarized and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog