Industry executives and experts share their predictions for 2024.  Read them in this 16th annual VMblog.com series exclusive.

Three Crucial AI Cybersecurity Predictions Ahead of 2024

By Tal Zamir, CTO of Perception Point

In 2023, a wave of remarkable AI tools and applications swept the tech landscape, putting once seemingly imaginary technologies into the hands of virtually all internet users. In 2024, even more extraordinary AI tools, applications and capabilities will emerge.

However, just like AI can be leveraged to scale productive workflows and bolster ROI, as with many other cutting-edge technologies, it also presents new opportunities for threat actors to orchestrate increasingly sophisticated cyberattacks.

In preparation for the challenges ahead, organizations must understand how threat actors can leverage AI in order to formulate comprehensive defense strategies and counter malicious attacks.

Hackers will let AI do the heavy lifting

For businesses, new and improved AI tools are meant to automate and therein alleviate much of employees' workloads, granting workers the capacity to perform a multitude of tasks in a fraction of the time it would have taken them to do manually.

Unfortunately, hackers are exploiting this very faculty.

By taking advantage of the very same AI tools used to enhance productivity at the workplace, hackers are customizing ChatGPT-like bots in alarming ways. These customized bots can be used as social engineering powerhouses - they can be designed to create highly personalized messages and carry out targeted phishing campaigns to gain access to sensitive company data such as private customer information, financials, and intellectual property.

Moreover, attackers can optimize their operations with the help of AI and Large Language Models (LLMs). Through AI-based automation, malicious actors can launch large-scale social engineering campaigns, take over user email accounts, hijack email threads to trick victims into transferring funds, and move laterally across the organization in a viral way.

AI's ability to automate the preliminary grunt work needed to fashion sophisticated attacks and become weaponized for a wide range of malicious acts is a testament not only to its versatility, but also its potency in crafting sophisticated attacks without significant effort. Organizations of all sizes across various industries should ensure they upgrade their defenses across the main communication channels, including email, browsers, and collaboration apps.

AI will make attacks harder to detect

Beyond its ability to amplify both the volume and frequency of cyberattacks, AI has granted hackers the ability to engineer and execute personalized attacks with heightened sophistication.

In particular, the widespread availability of multimodal machine learning models has empowered hackers to fabricate and deploy shockingly realistic deepfakes - including image doctoring and audio-visual impersonation - to scam employees into giving up sensitive company information. 

These techniques raise the effectiveness of social engineering attempts to new heights, whereby a specific individual's characteristics or mannerisms are replicated so convincingly that the average user has great difficulty detecting that they are forged.

When targeting an individual employee who normally receives a barrage of work emails on a daily basis, for example, the chances that one malicious email goes undetected is high, and are greatly affected by the efficacy of the deployed security systems. Without sufficient due diligence and implementation of effective automated security controls and security awareness training to identify malicious content, employees won't be able to exercise the necessary caution. 

Software companies will face off against adversarial prompts

Many tech companies are now building their own AI-powered tools to provide advanced GenAI-based services to their customers.

Guarding against malicious and adversarial prompts in LLMs will be a cornerstone of securing digital assets across industries in the coming years. Hackers are already manipulating or outright disabling the safeguards built into LLMs that prevent generative AI systems from dispensing criminal content.

In the coming year, a cohort of startups committed to safeguarding against adversarial prompts and providing cutting-edge solutions to mitigate the risks such threats pose will likely arise. By addressing the elevated risk of this attack category, organizations can maintain a competitive edge throughout 2024.

AI vs AI

Hackers will continue leveraging AI to launch sophisticated attacks more frequently and more precisely as its capabilities grow ever more potent. As AI-driven attacks have proliferated within the threat landscape, organizations have already been advancing their own AI defenses to thwart these mounting threats and must continue to improve upon these tools as attackers simultaneously increase the sophistication and scope of their own AI attacks.

Using AI defensively will continue to be a critical component of cybersecurity in 2024, especially to counter these newly emerging threats.

##

ABOUT THE AUTHOR

Tal is the Chief Technology Officer at Perception Point. Previously the Founder & CTO of Hysolate, Tal Zamir is a 20-year software industry leader with a track record of solving urgent business challenges by reimagining how technology works.

He has pioneered multiple breakthrough cybersecurity and virtualization products. Tal incubated next-gen end-user computing products while in the CTO office at VMware.