How Businesses Can Balance AI Innovation and Risk

Artificial intelligence is reshaping businesses at rapid speeds, but one factor threatens to slow innovation: trust. Or, more accurately, the lack of it.

“Trust is the foundation of innovation in AI, and without it, businesses hesitate to fully unlock AI’s potential,” says John Scimone, President and Chief Security Officer for Dell Technologies. 

The last year brought the transformative emergence of agentic AI—autonomous systems capable of solving complex problems without direct human oversight. While these systems can drive unprecedented efficiency and innovation, they also amplify concerns around accountability, ethical decision-making and security. 

These concerns span the entire AI infrastructure—from hardware to software stacks. The risks are accelerating: a recent Synopsys report shows a 236% increase in high-risk attack patterns in open-source software vulnerabilities across AI and data technologies over the past five years.

Earning trust in AI means going beyond security to include proper governance, ongoing monitoring and clear accountability. As AI advances to autonomous agentic systems, enterprises must recalibrate their security approaches to ensure end-to-end protection across increasingly complex AI workflows. While an ISV may provide data protection from external threats, governance is required to provide the quality, accuracy and relevance of data used by agents.

Organizations that fail to establish comprehensive oversight frameworks don't merely face operational risks—they undermine their competitive position and broader societal trust.

“These risks are barriers to innovation,” says Scimone. “Organizations have a responsibility to protect not only their businesses but also the people and systems their AI applications impact.”

For businesses, trust begins with confidence that their devices and infrastructure are secure from desktop to data center. As building this trust becomes more complex with agentic AI workloads, the Dell AI Factory with NVIDIA addresses the new autonomous AI challenges. This solution leverages NVIDIA’s approach to securing AI factories through embedding security features into every layer designed for AI workloads.

NVIDIA approaches the rising challenges through key principles, starting with Zero-Trust Architecture, which assumes no implicit trust and isolates applications from underlying infrastructure while continuously monitoring for threats in real time. This principle is realized through the NVIDIA Morpheus cybersecurity AI framework, which performs real-time AI inference on streaming data. This allows security teams to detect and respond to threats as they emerge—a critical capability for fast-moving agentic AI environments.

Other key NVIDIA security principles include secure software by design, which integrates robust security practices throughout the entire product lifecycle; proactive and continuous assessment through practices like LLM red teaming, which systematically tests autonomous AI models and discovers novel vulnerabilities before deployment; and separation of concerns, which distinguishes the complexity of an AI’s workflow from the risk of the tools it uses. That means developers can build innovative agentic AI systems while granularly managing the risks associated with specific functions. 

To implement these security principles, the partnership combines the security capabilities of NVIDIA accelerated computing and AI software with Dell’s secure hardware foundation. Dell offers customers secure commercial Dell Precision workstations built with security baked into the deepest levels to protect devices across their lifecycle. They leverage neural processing units to offer on-device AI for faster, real-time threat detection.

Dell Trusted Device software provides telemetry for SafeBIOS features like threat detection and BIOS verification, integrating with existing security ecosystems, while organizations can deploy Dell hardware with reference architectures for encrypted data in transit and at rest, plus NVIDIA GPU-enabled confidential computing for encryption in use — essential capabilities for workstations deploying autonomous AI applications.

Building on the hardware layer, Scimone says there's one powerful approach: Using AI to protect AI. Dell's endpoint protection solutions leverage behavioral analytics and advanced threat detection designed to identify anomalous behavior in autonomous AI applications.

End-to-end protection extends through Dell PowerScale storage solutions for secure, scalable data management, while Dell PowerEdge servers integrate security directly into the infrastructure.

The edge is particularly vulnerable, Scimone says, because it operates outside traditional data center boundaries, creating challenges with physical access control, expanded attack surfaces, and difficulties in updating, monitoring and managing distributed systems.

Dell's edge solutions are equipped with built-in security protocols, such as threat monitoring and encrypted communication channels, designed to address these unique edge challenges and secure data where it's generated.

"This ensures sensitive data remains protected from edge to core to cloud," says Scimone.

Agentic AI is rewriting the security playbook. Traditional security frameworks that worked for human-supervised AI can break down with systems that operate independently.

As businesses move toward adopting agentic AI to unlock new possibilities, they must also address the heightened risks these systems bring. By prioritizing security from day one, embedding it at every layer and implementing robust governance controls, organizations can ensure that even autonomous AI systems operate responsibly and securely. The Dell AI Factory with NVIDIA provides the secure foundation organizations need to navigate these new challenges.

AI is amplifying the importance of data, but this also magnifies the risks. When data falls into the wrong hands, it can lead to breaches, exploitation and serious consequences that threaten privacy and security. To mitigate these risks, organizations can turn to on-premises deployments that give them greater control over their data. The Dell AI Factory with NVIDIA approach can be up to 62% more cost effective for on-premises AI deployments than public cloud alternatives, while enabling organizations to maintain control over their governance and security posture, according to an Enterprise Strategy Group white paper commissioned by Dell.

The organizations that succeed will be those that recognize AI trust as requiring both secure technological foundations and robust governance capabilities. By pairing cutting-edge security technologies with rigorous oversight frameworks, organizations can confidently unlock AI's transformative potential — while maintaining trust.