Artificial Intelligence & Machine Learning , Next-Generation Technologies & Secure Development

Firms Eye Gen AI for Cybersecurity but Expect Little Impact

In a year in which the financial impact of cyberattacks has more than doubled to $1.4 million, organizations are exploring generative artificial intelligence to improve their cybersecurity posture, Dell said in a Tuesday report.

See Also: On-Demand | Security Operations In a New Paradigm

A majority of companies are concerned that their organization's existing data protection measures are unable to cope with malware and ransomware threats, according to the company's nonscientific survey of 1,500 IT and security decision-makers.

Nearly half of the respondents from the Americas, China, Europe, Asia-Pacific, Japan and the Middle East and Africa regions said their companies are ready to use machine learning and analytics to ensure that their backup data is "clean" before recovering it in the event of a cyberattack.

"Very few" vendors are putting gen AI capabilities into their security products, and they lean more toward machine learning rather than AI, especially in the anomaly detection space, said Rob Emsley, director of product marketing for data protection at Dell.

Generative AI's ability to analyze vast amounts of data quickly can identify abnormal patterns or behaviors that might indicate a cybersecurity threat. It can detect anomalies in network traffic or data patterns, unusual user behavior, or suspicious email content, enabling quicker identification of threats such as malware, phishing attempts or insider attacks, Emsley said.

But automation is where AI will have the "biggest impact" in the very near term as vendors come up to speed on how AI can help the existing solution provide benefits, he told Information Security Media Group. It can help automate routine cybersecurity tasks, such as patch management, network monitoring and incident reporting. "This not only increases efficiency but also frees up human security personnel to focus on more strategic and complex tasks," Emsley said.

Enterprises can also use artificial intelligence to prepare for security incidents by using its predictive capabilities. AI can learn from past incidents and global cybersecurity trends to predict future threats or identify potential vulnerabilities. It can also add capabilities to standard penetration testing to advance its capabilities.

Gen AI represents a "turning point" for cybersecurity, both as a challenge and a solution, Emsley said. Still, survey respondents appear to have low expectations from the new technology for their long-term cyber resiliency plans.

Slightly more than half the respondents say generative AI will only initially provide organizations an advantage in defending themselves against cybercriminals. An average of 20% of respondents believe that the technology will assist both the attackers and the defenders.

Cybercriminals are likely to take advantage of gen AI in the near term to circumvent cyber defenses, Emsley said, citing Dell Chief Technology Officer John Roese.

The degree of adoption of generative AI in cyberattacks will vary depending on the sophistication of cybercriminal groups, their resources, and the specific attack vectors they choose. Bad actors have so far used the technology to carry out more realistic phishing attacks and voicemail spoofs and to generate deepfake videos, he said.

Nearly all of the 1,500 respondents - 88% - said the emerging technology is likely to generate "large volumes of new data," and that it will increase the value of certain data types that a majority agree may require higher data protection service levels. Most respondents also expect to deploy gen AI on-premises and in the public cloud, Emsley said.

Addressing data protection concerns for generative AI requires similar foundational strategies to those used for other data sources and applications, whether in a multi-cloud or on-premises environment, he said. "This begins with a comprehensive understanding of how an organization plans to deploy and utilize gen AI for their business, encompassing the protection needs of input data, processing and output," he said.

The concern is not specific to generative AI, as nearly all organizations face data protection challenges, and 52% of the surveyed organizations say they have experienced "significant" disruption over the past 12 months, marking the highest percentage in over five years. The companies lost 2.17 terabytes worth of data in the period, and damages and recovery cost them $2.61 million. The cyber-related incidents themselves cost organizations about $1.92 million.

Seventy-nine percent of respondents are concerned about experiencing a disruptive event in the next year.

Three-fourths of the participating organizations are worried that their existing data protection measures may not effectively handle ransomware threats, while only 59% said they invest more in cyber prevention than cyber recovery, highlighting a potential imbalance in cyber resilience strategies. Many companies - $42 - appear to have invested an undisclosed amount of funds and efforts to use AI to "better defend" against attacks, but the survey does not include details about these investments.