SPARK Insights

AI Can Make Your Company More Efficient. It Can Make Your Adversaries More Efficient, Too.

By Jerry E. Beckley II, Founder, Spark Infrastructure Advisors, LLC

October 2026

Artificial intelligence is forcing company leaders to confront two possibilities at once: better performance and greater exposure.

We see promises of faster work, lower costs, and better decisions alongside headlines about job losses, misuse, and systems behaving in unexpected ways. Some describe AI as our salvation. Others see it as our undoing.

My view is that leaders should approach AI with curiosity, discipline, and a clear understanding of consequences. The question is whether we are preparing our organizations as carefully as we are pursuing the benefits.

A pledge is a starting point

The Associated Press recently reported that major AI companies signed a White House agreement on AI safety amid growing debate over industry self-policing.^[1] Cooperation matters. Developers should test their systems, explain limitations, and take responsibility for harm.

But a public commitment cannot answer the questions inside your organization. What information are employees entering into AI tools? Which systems can those tools access? Who approves consequential actions? What happens when an output is wrong or a service becomes unavailable?

A vendor’s commitment is one part of the picture. Leadership still owns the decision to use the technology and the consequences of that decision.

Efficiency works both ways

AI can help teams draft communications, summarize documents, analyze information, and reduce repetitive work. Those are worthwhile opportunities when results are checked and sensitive information is protected.

Adversaries can pursue efficiencies, too.

The United Kingdom’s National Cyber Security Centre assesses that AI will almost certainly make elements of cyber intrusion more effective and efficient. Its assessment identifies uses including reconnaissance, social engineering, vulnerability research, and basic malware generation.^[2]

For a business leader, the implication is straightforward: attackers may be able to research your organization, develop convincing approaches, and exploit weaknesses faster.

Consider a hypothetical payment request that appears to come from a familiar executive. It references a real project and pressures an employee to act quickly. The defense needs to include an independent verification process that employees are expected—and empowered—to follow.

Similarly, connecting an AI assistant to company records or business applications deserves careful review. An assistant that summarizes information carries different consequences from one authorized to change records or initiate transactions. Give each tool only the access and authority its purpose requires.

Mistakes matter even without an attacker

An organization can suffer harm through ordinary, well-intentioned AI use.

The National Institute of Standards and Technology identifies confidently presented false information and excessive human reliance on AI among the risks of generative AI.^[3] A polished answer can still be wrong.

Imagine an inaccurate summary influencing a contract decision, or an employee entering confidential information into an unapproved tool. These are reasons to define acceptable use, review outputs according to their consequences, and assign a person accountable for the result.

Human review needs enough time, expertise, and authority to challenge the system. Otherwise, it becomes a signature on a decision nobody meaningfully examined.

Prepare people as carefully as technology

Concern about employment deserves an honest response.

The International Labour Organization’s 2025 analysis estimated that roughly one in four jobs worldwide had some potential exposure to generative AI. It concluded that job transformation was more likely than complete replacement because human input remains necessary across many tasks. Exposure is not a count of jobs already lost.^[4]

That finding does not remove the uncertainty employees feel. Leaders should explain what they intend to change, involve the people doing the work, and invest in training.

Before treating time saved as a reason to reduce staffing, ask how that capacity could improve service, strengthen controls, or address neglected work. Also ask whether your organization is preserving the knowledge it will need when automation fails.

Turn concern into decisions

Leaders do not need certainty about AI’s ultimate destination to take useful action now.

Start by identifying where AI is already being used, including features inside existing software. Assign an accountable owner. Set rules for approved tools, sensitive information, access, and human approval. Evaluate benefits alongside errors, rework, and new dependencies.

Then rehearse a disruption. Bring leadership, operations, technology, finance, and communications into the same conversation. Work through a fraudulent request, an incorrect automated action, or the loss of an important provider. Decide who can stop the affected process, how essential services will continue, and what customers need to hear.

The exercise should produce decisions, owners, and improvements that somebody follows through on.

Where Spark can help

Spark Infrastructure Advisors helps leaders turn uncertainty into practical readiness.

That work connects policies, leadership responsibilities, incident response, disaster recovery, and operational priorities. Through readiness assessments, policy reviews, and facilitated exercises, Spark can help your organization examine its dependencies, clarify decisions, and identify gaps before a disruption exposes them.

For AI, the goal is to pursue useful capabilities with an organization prepared to manage errors, misuse, and interruption.

A sensible first step is a leadership discussion about what your business must keep doing—and whether your current plans support it. Start at sparkcritical.com/readiness.

Readiness is not predicting what will go wrong. It’s knowing what you will do when something does.

References

1. Associated Press. In Silicon Valley, AI founders cheer Trump’s calls for the industry to police itself on safety. Reporting accessed October 10, 2026.

2. UK National Cyber Security Centre. Impact of AI on cyber threat from now to 2027. May 7, 2025.

3. National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. July 2024, sections 2.2 and 2.7.

4. International Labour Organization. How might generative AI impact different occupations?. May 20, 2025.