WJ Talk: Maritime companies should apply ‘bridge discipline’ to AI governance, attorney says

As artificial intelligence becomes more common across maritime operations and business functions, companies should approach AI governance much like they approach vessel operations, according to Michelle Ramsden, special counsel at Jones Walker.

Speaking during a recent Waterways Journal Talks interview, Ramsden said maritime organizations can draw lessons from bridge resource management, where clearly defined responsibilities, communication protocols and escalation procedures help prevent operational failures.

Ramsden, whose practice focuses on commercial transactions, privacy, cybersecurity and artificial intelligence, said organizations should resist the temptation to treat AI systems as autonomous decision-makers.

“You can’t turn over accountability to technology,” Ramsden said. “There still have to be people responsible for ensuring these systems are implemented and used appropriately.”

She said AI-related problems often emerge gradually before escalating into larger operational, financial or reputational issues. As a result, organizations need governance structures that clearly identify who is responsible for oversight and decision-making.

Ramsden pointed to the National Institute of Standards and Technology’s AI Risk Management Framework as one approach organizations can use to establish those guardrails. The framework centers on four functions: govern, map, measure and manage.

The governance phase focuses on establishing policies, accountability structures and organizational objectives for AI use. Mapping involves identifying where AI tools exist within an organization, how they are being used and what systems or data they affect.

The measurement stage evaluates performance, reliability, bias and potential risks associated with AI tools. Management involves ongoing oversight, monitoring, corrective actions and incident response.

According to Ramsden, many companies struggle simply because they do not have a complete picture of where AI tools are being used across their organizations.

“It’s difficult to govern something if you don’t know it exists,” she said.

The conversation also addressed the growing importance of vendor contracts as organizations adopt third-party AI platforms.

Ramsden said companies should pay close attention to how vendors handle proprietary data, whether information submitted to AI systems could be incorporated into training models and how liability is allocated if problems arise.

She noted that AI can magnify risks more rapidly than many traditional technologies because decisions and recommendations often build upon previous outputs.

“A minor error can become a major issue quickly,” she said.

The maritime sector is not immune to those risks. Ramsden cited examples of falsified AIS signals that caused vessels to appear in locations where they were not actually operating, demonstrating the importance of verifying digital information before acting on it.

At the same time, she said AI can provide significant benefits when used appropriately.

Organizations can use AI to improve efficiency, increase operational scale, streamline documentation reviews and identify unusual patterns or changes that may indicate emerging risks. AI-powered monitoring tools can also help detect performance drift in systems before larger issues develop.

Ramsden cautioned, however, that organizations should not feel pressured to fully solve AI governance before experimenting with the technology.

Instead, she encouraged companies to start with limited use cases, learn from those experiences and gradually expand their capabilities while building governance structures alongside adoption.

The discussion also touched on workforce development and the growing importance of AI literacy.

While certifications and specialized training programs are becoming available, Ramsden said AI knowledge is increasingly becoming a foundational skill that can be applied across professions, including maritime operations, law, communications and healthcare.

Understanding the context in which AI is being used remains critical, she said.

“It’s not enough to know AI. You also have to understand the industry where you’re applying it,” Ramsden said.

Looking ahead, Ramsden expects AI to become increasingly embedded in everyday technology and business operations. Success, she said, will depend less on having the newest tools and more on having disciplined processes governing their use.

“AI is a tool,” Ramsden said. “The organizations that are successful won’t just have better tools. They’ll have better discipline around how those tools are used.”