As the UAE pushes forward with its ambitious plan to integrate AI agents across 50 percent of its public sector, cybersecurity leaders are calling for caution. While the potential for efficiency is high, recent incidents—such as the breach involving OpenAI’s models at Hugging Face—have highlighted the dangerous reality of AI "going rogue." Experts argue that these risks are not inevitable, provided that organizations implement robust foundational security controls rather than treating AI as an untouchable black box. According to Aleksandar Valjarevic of HelpAG, the key is to move away from using frontier models for every task and to ensure that human oversight remains the final word in every operation.

The challenge is intensified by the "novelty gap," a critical window where cybercriminals leverage machine-speed attacks before traditional defense systems can catch up. Kent Breaux of Proofpoint notes that attackers currently hold a significant advantage, potentially deploying exploits in mere seconds while security firms scramble to patch vulnerabilities. Despite the daunting speed of these AI-driven threats, the consensus among industry professionals is that humanity still holds the upper hand. By strictly defining how these tools are utilized and maintaining a culture of accountability, the UAE can successfully harness the power of agentic AI without sacrificing its digital security.