.webp)
As 2025 approaches, the landscape of AI in the enterprise is undergoing significant shifts, with a focus on practical applications rather than the grand disruptions once anticipated. Norman Rice, Chief Commercial Officer at Extreme Networks, suggests that many organisations initially viewed AI as a tool for widespread automation. However, as the frenzy subsides, companies are looking to leverage the technology for incremental efficiency improvements rather than wholesale changes. This shift mirrors AI's transition into the 'trough of disillusionment' phase of its hype cycle, prompting firms to pursue more ROI-driven uses, particularly in networking and security. AI is expected to enhance productivity by facilitating faster certification for specific technologies and improving real-time IT issue detection.
The potential for AI as a tool for cybercrime is a growing concern for industry leaders. Mark Bowling, Chief Information Security and Risk Officer at ExtraHop, warns of an increase in traditional fraud facilitated by generative AI, which is now within the reach of hackers. The technology enables sophisticated impersonation tactics posing as police officers or executives to illicitly acquire credentials and personal information. In response, emphasis on identity protection is expected to increase, with stronger authentication methods like multi-factor authentication (MFA) and Identity Access Management (IAM) tools playing critical roles.
AI's influence on trust is also at the forefront of discussions. Andre Durand, CEO and Founder of Ping Identity, anticipates that AI-generated impersonations will necessitate a 'trust nothing, verify everything' culture. In this evolving landscape, verification will supplant authentication as the cornerstone of trust, urging individuals and organisations to rely only on verified information.
The role of AI in cybercriminal activities is set to escalate as threat actors use it to generate highly personalised phishing attacks and adaptive malware, according to Sadiq Iqbal, Chief Security Advisor at Check Point Software Technologies. This evolution will allow less skilled hackers to perform sophisticated operations, thereby democratising cybercrime further.
As the industry grapples with AI's practical applications, Morey Haber, Chief Security Advisor at BeyondTrust, predicts the de-escalation of AI's inflated promises. While AI will remain integral for basic automation and workflows, the reality of its capabilities, especially in security, will fall short of the hyperbolic claims made. The focus is expected to shift to narrow AI applications designed for practical security and workflow enhancements, reducing operational complexity and skill needs.
According to Corey Nachreiner of WatchGuard Technologies, by 2025, multimodal AI's integration will enhance the efficiency and sophistication of cyberattacks. These systems, capable of handling text, images, and voice, will empower threat actors to streamline attack pipelines, automating processes from phishing to malware deployment. This evolution will present significant challenges for organisations and security teams, requiring heightened vigilance and response strategies.
The introduction of a 'Zero Trust for AI' mandate is proposed by Steve Povolny, Senior Director at Exabeam, cautioning against the blind reliance on AI outputs. This concept advocates for rigorous verification of AI-generated data before it informs critical security decisions, fostering a culture of meticulous scrutiny in AI deployments. Human oversight is projected to become an essential component of AI integration in security to mitigate risks associated with unchecked AI outputs.
The year ahead is poised to see AI's impact on business and security environments shift towards more grounded, focused applications, as organisations learn to navigate both its potentials and threats more pragmatically. This evolution will demand renewed emphasis on verifying and substantiating AI-driven operations, underlining the importance of human oversight in the integration of AI technologies.
