Over the past two weeks, we’ve reflected on 2025’s cybersecurity developments—with a few glimpses into 2026. Today, we turn our full attention to the year ahead, sharing predictions from leading experts on what’s next for cybersecurity.
AI will elevate phishing and identity-based threats
Brian Keeter, senior director at APCO and senior fellow at the McCrary Institute for Cyber and Critical Infrastructure Security, warns that AI will make phishing attempts more sophisticated and convincing, leading to more victims.
He predicts identity theft will become a dominant threat. While network security remains critical, attackers will increasingly target individuals using tools that make scams appear credible. Organizations lacking updated incident response and stakeholder engagement plans risk losing trust and confidence.
Architectural risks in AI adoption
Greg Sullivan, founding partner of CIOSO Global, highlights a less-discussed risk: the architectural shortcomings of AI tools as companies rush to scale adoption.
Current large language models (LLMs) lack persistence and reasoning, treating each task as new without reliable memory or context. This gap creates inconsistency, compliance drift, and repeatable vulnerabilities at scale. While retrieval-augmented generation and prompt libraries help, they are superficial fixes that won’t scale cleanly.
Sullivan emphasizes that success in AI still depends on “boring” fundamentals—data governance, management, and ownership. In 2026, most AI-related failures will stem from poorly governed data pipelines rather than the models themselves.
Old threats persist, compliance becomes key
Cam Roberson, VP at Beachhead Solutions, notes that 2026 won’t be defined by new threats but by the persistence of old ones—phishing, credential theft, and insider errors. Many SMBs still rely on perimeter defenses alone, leaving gaps MSPs can address.
Progress will come from layered security becoming standard rather than premium—encryption, antivirus, and access controls that mitigate ransomware 2.0 attacks focused on data theft and extortion. Compliance frameworks strengthened in 2025 will create both obligations and opportunities for MSPs to deliver scalable, repeatable, compliance-ready security offerings.
Agentic AI and the need for governance
Chirag Shah, Global Information Security Officer & DPO at Model N, predicts traditional mass-scale attacks will decline as agentic AI autonomously maps vulnerabilities, enabling faster breaches. Personalized phishing campaigns will bypass skepticism and traditional defenses.
Speed will define success for defenders, requiring automated containment and real-time policy enforcement. Shah stresses the need for AI governance committees, clear policies, and expanded CISO responsibilities. Organizations that fail to invest in governance risk exposing valuable AI models to misuse and regulatory fallout.
Cybersecurity in 2026 will be shaped by AI—both as a tool for attackers and a challenge for defenders. Organizations that prioritize governance, layered security, and proactive planning will be best positioned to navigate the evolving threat landscape.
Photo: MHZZ / Shutterstock
This post originally appeared on Smarter MSP.