Australia warns of ‘unprecedented’ AI threat

The Australian Signals Directorate has issued a warning about the unprecedented threat AI poses as foreign governments exploit it to hack computer networks and critical infrastructure. The cybersecurity agency reported a staggering 80% increase in potentially malicious activity over the past year, with AI playing a significant role.

While AI can be harnessed for good, it has also empowered criminals and state-sponsored hackers to launch attacks at unprecedented scale, speed, and sophistication. The report highlights the rapid advancement of AI models, which can discover vulnerabilities more quickly and exploit them with greater effectiveness. In recent months, new threats have emerged from unintended actions of AI agents and model testing.

Global concerns about advanced AI tools intensify after a series of hacking incidents involving models from leading developers like OpenAI and Anthropic. The report documented a surge in cybersecurity incidents, with state-sponsored actors primarily targeting Australian organizations to gain access to sensitive information, affecting government networks, critical infrastructure, private businesses, and individuals.

Defending Minister Richard Marles emphasized that AI is a real and present reality, urging Australians to update their systems and harden their defenses against malicious actors using AI to find vulnerabilities. The urgent need to address this emerging threat has prompted governments and firms to quickly patch critical flaws in programming languages, particularly those running a large portion of the world's websites, which were found using a powerful OpenAI model.