Limited reforms could allow the work of biological computers to be recognised by copyright law, study shows

The work of neurons acting as biological computers can be protected by copyright law with “limited” legal reforms, new research shows.

The artificial brains are increasingly used for medical research, computer software, art and even quantum computing, but current law ultimately requires human authorial involvement, whether through the exercise of skill, labour and judgement, creativity, or the arrangements made for computer-generated works.

The study examines why neurons themselves do not qualify as copyright works, and why autonomous neuronal activity presents difficulties for conventional concepts of authorship.

However copyright law could recognise the computer coding which controls the neuron, as long as this was demonstrated in a design document. This approach will bring greater legal certainty.

Professor James Griffin, from the University of Exeter Law School, who carried out the analysis, said: “Traditional copyright assumes that a work is the product of a mind that remains external to it, biological computing collapses that separation.

“Neurons are increasingly used for important tasks such as medical research, computer software, art and even quantum computing, so this is a gap in the law which needs to be filled. Copyright needs to be available for work completed by biological computers.

“We cannot copyright neurons, but we could copyright the code around them.”

Professor Griffin’s suggested reform would apply when living neuronal cultures are embedded within software and other copyright works as critical functional components.

Neurons are the basic functional units of the nervous system. They receive, process, and transmit information through electrochemical signals. A typical neuron consists of a cell body (soma), dendrites that receive input from other neurons, and an axon that carries the output signal onwards.

Neurons can be derived from something as readily accessible as skin cells. These cells may be reprogrammed into induced pluripotent stem cells (iPSCs), which can subsequently be differentiated into neurons. The resulting cells are then plated onto a multi-electrode array dish, treated to encourage adhesion and network growth, and maintained in a controlled incubator environment (temperature, CO2, humidity, nutrient medium) while they extend neurites, form synapses, and may develop spontaneous electrical activity, including spikes and co-ordinated network bursts, which can be detected by the electrodes.”

Under UK law, protection subsists only in literary, dramatic or musical works that are “recorded, in writing or otherwise”, such that the work has some material form.

Professor Griffin said: “Neuron based systems reveal a contradiction within UK copyright: the copyright attaches to the digital shell, but the behaviour that gives the work its distinctive character is realised through a biological process that resists classification as a literary work at all, and sits closer to the domain of unprotectable ideas, methods, or operations than to authored expression.

“Rather than a blanket statutory exclusion of ‘biological matter’ from copyright protection, a more precise reform is preferable. Neurons in their natural or passively responsive state remain unprotectable. However, where developers embed a neuron culture within a structured computational architecture, subject it to a deliberately designed training curriculum and reward regime and stabilise the resulting behaviour such that it can be characterised and deployed as a functional controller within a design document, the law should be open to recognising subsistence in configurations of coded and trained neurons.”

The analysis says neurons which are not trained or coded would not be protectable under current law, and that this should remain so. Instead, protection is to exist for a configuration of coded and trained neurons specifically where humans embed the culture within a structured architecture; subject it to a deliberately designed training curriculum and reward regime and stabilise the resulting behaviour.

The design document would require the developer to fix, in material form, the elements that define the neuron coding and training. This could include the interface mappings, training curriculum, stimulation and reward protocols, constraints, and performance benchmarks.