SandboxAQ, a leader in AI and quantum technology solutions, released its first AI Security Benchmark Report on July 24, 2025, highlighting a critical disconnect in enterprise cybersecurity. While 79% of organizations have adopted AI in production environments, only 6% have implemented AI-native security strategies, leaving most vulnerable to threats like model manipulation, data leakage, and adversarial attacks. Based on a survey of over 100 senior security leaders across the US and EU, the report underscores the urgent need for modernized security approaches to address the evolving AI-driven threat landscape.
79% of enterprises use AI, but only 6% have AI-native security.
74% of security leaders fear AI-enhanced cyberattacks.
69% concerned about AI exposing new vulnerabilities.
Only 10% of organizations have dedicated AI security teams.
Non-human identities (NHIs) complicate security governance.
85% plan increased AI security spending within 12-24 months.
The rapid integration of AI into enterprise operations has outpaced cybersecurity readiness. The report finds that 79% of organizations are using AI in production, yet only 6% have comprehensive AI-native security measures. “This isn’t just a solution gap, it’s a conceptual one,” said Marc Manzano, General Manager of the Cybersecurity Group at SandboxAQ. “AI is radically changing the cybersecurity paradigm at an unprecedented speed.” With 74% of security leaders highly concerned about AI-enhanced cyberattacks and 69% worried about new vulnerabilities, the need for specialized AI security is evident.
Non-human identities (NHIs), such as autonomous AI agents and machine accounts, are a growing concern. These systems operate independently, accessing sensitive resources without human oversight, yet most organizations lack visibility or control over them. Only 10% of companies have dedicated AI security teams, leaving traditional IT and security teams to manage these complex entities. This gap undermines Zero Trust principles and exposes weaknesses in identity governance and cryptographic hygiene, increasing the risk of undetected threats.
The report highlights that only 28% of organizations have conducted AI-specific security assessments, with most relying on traditional, rule-based tools ill-suited for dynamic AI systems. These legacy solutions struggle to address threats like model manipulation or data leakage at machine speed. SandboxAQ’s AQtive Guard offers a solution by modernizing cryptographic and identity governance, enabling enterprises to extend core security practices like automated inventory and policy enforcement to AI infrastructure.
Despite current gaps, enterprises are prioritizing AI security, with 85% planning to increase spending over the next 12-24 months. Key focus areas include protecting training data and inference pipelines, securing NHIs, and deploying automated incident response tailored to AI systems. This shift reflects growing recognition among CISOs that defending against AI-driven threats requires new, proactive approaches rather than layered patches on outdated tools.
SandboxAQ’s AI Security Benchmark Report underscores the urgent need for enterprises to align cybersecurity with AI adoption. By highlighting vulnerabilities and advocating for AI-native security solutions like AQtive Guard, the report positions SandboxAQ as a leader in addressing the evolving challenges of AI-driven enterprise environments, paving the way for more secure and resilient AI implementations.
SandboxAQ is a B2B company delivering solutions at the intersection of AI and quantum techniques. The company's Large Quantitative Models (LQMs) deliver critical advances in life sciences, financial services, navigation, and other sectors. The company emerged from Alphabet Inc. as an independent, growth-backed company funded by leading investors including funds and accounts advised by T. Rowe Price Associates, Inc., IQT, US Innovative Technology Fund, S32, Hillspire Capital, Breyer Capital, Marc Benioff, Thomas Tull, Paladin Capital Group, and others.