The IP of the data feed into ML for AI to be used in EM conundrum / by Francisco Blaha

Besides being one of the most ridiculous blog titles I’ve ever written… I think we are in another one of those situations that we wish we had paid attention to earlier.

I’ve been writing about EM for a while now (2016), yet I've never paid much attention to the issue until a conversation with a friend brought it up.

What are we talking about?

Electronic monitoring (EM) is an increasingly important tool for enhancing the management of Pacific longline fisheries. Although it has costs and limitations, EM offers indisputable benefits, such as improved scientific coverage, transparency, and compliance.

As EM initiatives and trials progress in the Pacific Island Countries and Territories (PICTs), one important but sometimes neglected question arises: Who gets to keep the IP from the data produced by these trials?

This is not a simple question. It is common practice for service providers to maintain ownership of critical intellectual property when EM systems are deployed, especially in the software used to analyse EM imagery, regardless of whether the deployment is part of a government-led program or an industry-driven pilot.

This category includes Machine Learning (ML) that feeds into the broader concept of Artificial Intelligence (AI) systems that can detect gear use, automatically identify species, or flag possible compliance problems. Like how an apprentice gains intelligence through experience, these tools get smarter with each dataset they consume.

The concern is this: if a service provider is improving their product using data and experience gained from operations in Pacific waters, without appropriate legal agreements, PICTs may unwittingly give away valuable IP (a non-tangible asset until needed) that feeds the proprietary software of the EM providers to enhance their systems, which the EM providers will then charge PICTs to use.

It is somewhat like a Mexican standoff… EM software providers need the dataset for their product to be helpful PICTs... And PICTs need the software to run the EM systems.

The EM service provider benefits immensely (from refined algorithms, better software and stronger commercial positioning). At the same time, the country receives limited long-term value beyond the initial project goals.

This is particularly true because in most EM trials in the region, costs and equipment are paid by NGOs with the support of willing industries, yet the final users (and payers) of the system will be the PICTS themselves and any other coastal states in the world that later get a subscription to the EM software.

The EM service provider benefits immensely (from refined algorithms, better software and stronger commercial positioning) while the coastal states receive limited long-term value beyond the initial project goals.

This has nothing to do with reversing advancement or questioning the worth of EM or the function of the private sector. It ensures Pacific nations get long-term advantages, equity, and sovereignty.

Pacific fisheries data and real-world conditions are becoming invaluable for training and refining AI as EM technology progresses. In the absence of appropriate agreements, PICTs are denying themselves a chance to claim ownership of the results of the data collected in their waters. In fact, UNCLOS Part XIII: Marine Scientific Research stipulates that any research in the EEZ and on the Continental Shelf shall be conducted with the coastal State's consent and delineates a series of conditions.

Regrettably, in EM, IP rights' contractual and legal components are frequently regarded as overly technical or ancillary. This is a mistake. Contracts with EM providers should contain provisions that safeguard national interests, particularly with data rights, benefit sharing, and intellectual property ownership.

Countries should proactively allocate legislative and legal resources to this end. Shared ownership of data-driven software enhancements, access to artificial intelligence models trained on regional fisheries, or even openness about how service providers use the data would be a good start.

 I’m sure this is a very complex area of negotiation, and involving lawyers would be expensive; however, ignoring this issue due to its complexity is counterproductive.

Potentially, a lot is at stake because EM is becoming more integrated into fisheries management; the older ones (like me) remember the lengthy negotiation process to get the PNA FIMS IP out of the developers.

Coastal States nations should be aware that letting ownership of critical technology slip away because of the unfamiliar legal terrain may be costly in the long term. Instead, they must see this as essential to safeguarding their sovereignty over natural resources and the digital tools underpinning their management structures.

As EM projects expand, coastal states must integrate the right legal frameworks and conditions on the EM trials contracts to ensure the benefits of any data-analysed innovation developed are shared fairly. Otherwise, coastal States risk being left behind. They may have to pay heavily for a revolutionary monitoring tool that they played a role in creating, but have no ownership or control over.