AI’s Double-Edged Sword in Cybersecurity
Artificial intelligence (AI) is poised to reshape the cybersecurity landscape in 2025, introducing both opportunities and significant risks. Experts predict that AI will amplify traditional threats such as phishing, insider attacks, and ransomware, adding layers of complexity. Generative AI (GenAI) in particular will supercharge cyberattacks, enabling threat actors to craft sophisticated phishing emails, create deepfakes ,Quantum Computing and and exploit unknown vulnerabilities.
Sanjeev Verma, co-founder of PreVeil, highlighted how GenAI enhances attackers’ capabilities, saying, “AI’s ability to process and analyze data will uncover vulnerabilities that organizations might overlook.” However, the increased reliance on AI also makes it a prime target. Adversaries are expected to manipulate large language models (LLMs) by contaminating private datasets, a tactic that could lead to harmful outcomes. According to Daniel Rapp, Chief AI and Data Officer at Proofpoint, securing AI-dependent systems will become a priority to prevent such manipulations.
Troy Bettencourt, head of IBM X-Force, emphasized the need for a proactive stance, distinguishing between AI-assisted and AI-powered threats. While AI-assisted attacks currently dominate, organizations must prepare for the evolution of AI-powered threats like deepfake scams and autonomous cyberattacks.
Quantum Computing and Multimodal AI Threats
As quantum computing matures, its potential for breaking current encryption methods looms large. Security experts warn that adversaries may soon exploit quantum capabilities to compromise encryption technologies, both existing and emerging. The implications could extend beyond virtual systems, with real-world consequences in sectors reliant on secure data transmission.
Multimodal AI, which integrates text, images, voice, and advanced coding, poses another significant challenge. Corey Nachreiner, CISO at WatchGuard, predicted that by 2025, attackers will utilize multimodal AI to automate entire cyberattack pipelines. This approach would democratize advanced cyberattacks, enabling even low-skilled threat actors to launch highly sophisticated and tailored operations.
The rising threats underscore the importance of adopting a “trust and verify” model for coding and security processes. Andrea Malagodi, CIO at Sonar, stressed that while AI tools enhance productivity, human oversight is indispensable. Without rigorous testing and quality assurance, AI-generated code risks introducing vulnerabilities that attackers could exploit.
Synthetic Identities and SaaS Security Challenges
The manipulation of synthetic content is expected to surge in 2025. Bad actors are likely to create realistic online personas using AI to influence public opinion, promote malicious agendas, or generate financial gains. Tyler Swinehart, Director of Global IT and Security at Ironscales, warned of fabricated experts gaining credibility through automated content creation. These synthetic identities could evade current screening technologies, spreading disinformation on a massive scale.
Meanwhile, the widespread adoption of software-as-a-service (SaaS) platforms introduces another layer of complexity. Threat actors could exploit SaaS document repositories and email systems to manipulate private data and undermine AI models. Security teams must prioritize safeguarding these platforms to counteract such threats effectively.
As the cybersecurity landscape becomes increasingly intertwined with AI and emerging technologies, organizations face the dual challenge of leveraging innovation while fortifying their defenses. Experts agree that vigilance, advanced tools, and human oversight will be critical in navigating the evolving threat landscape.
