Enterprise AI: Investment Up, Security Lags

Enterprise investments in AI are rising, with 86% of C-suite executives increasing their budgets. However, only 32% report a sustained impact from these investments. This gap shows enterprise AI is moving beyond pilot phases into operational challenges. New threats like prompt injection and shadow AI are contributing to these difficulties. Prompt injection, characterized as the functional equivalent of malware, allows attackers to embed instructions that cause AI models to act in unauthorized ways. 65% of organizations currently lack dedicated defenses against this type of attack. Security defenses have not kept pace with AI-specific attacks, and unsanctioned AI activity is running below the radar. Additionally, models are breaking due to underlying data issues, not faulty algorithms. This means organizations deploying AI without proper input validation and threat detection have significant security gaps.