84% of the US' critical infrastructure organizations have identified the use of AI to drive cyber threats as a current security concern.

This dramatic rise in concern about how cybercriminals use AI is revealed in new research by leading cybersecurity services firm Bridewell, surveying 519 staff responsible for cybersecurity in US critical infrastructure organizations, in sectors such as civil aviation, telecommunications, energy, transport, media, financial services and water supply.

The research found 87% of respondents are worried about AI-powered phishing attacks in which criminals use AI to radically improve the accuracy and wording of their email lures at scale. Criminals can also employ AI to complement basic coding skills, reducing the barrier to entry for exploits and enhancing the sophistication of their malware.

These developments are also why 86% of respondents voiced concerns about automated hacking. Meanwhile, 84% of respondents say they have fears about adaptive AI cyberattacks, that constantly evolve their tactics, and 85% expressed concerns about AI-driven exploit development.

All of the AI-driven threats listed in the research are of concern to more than three-quarters (75%) of respondents - including polymorphic malware which mutates with every infection. 80% said they fear this emerging threat.

The research also explored how critical infrastructure organizations are using AI to combat the increased use of AI by cybercriminal groups. AI-driven exploits or techniques are not yet as effective as conventional cyber tactics, and businesses are able to use AI-focused tools to protect their systems and infrastructure. With its ability to analyse large datasets rapidly, AI can be a useful tool in detecting malicious activity in a system or network, spotting anomalies and suspicious behaviour.

The research found current deployment of AI in cyber defences is in its early stages. Fewer than three-in-ten respondents' organizations are using AI-enhanced endpoint protection (29%), AI-driven data-loss prevention (28%), , or AI-based phishing detection and prevention (27%). Almost all organizations (98%) are, however, using some AI tools - a trend certain to gain momentum as cyber threats escalate and become even more sophisticated.

"While we are at the early stages of AI-driven cyber-attacks, concern among organizations is not unfounded as the technology presents itself as a future threat," said Chase Richardson ,  Vice President of Consulting at Bridewell. "Businesses can prepare for the impending AI arms race by incorporating the technology into their cyber defence strategies. AI can be a force for good by helping critical infrastructure organizations to enhance threat intelligence capabilities and accelerate detection and response strategies." 

Download the report.