Coding with AI: Navigating copyright in the era of Copilot
In the age of widespread artificial intelligence (AI) integration, over 90% of programmers now utilize AI tools such as ChatGPT, Microsoft Copilot, or Github Copilot in their software development workflows. Simultaneously, legal debates persist regarding the copyright implications related to content produced using AI, which has significant ramifications. Determining whether software elements created using AI qualify for copyright protection profoundly impacts licensing and copyright transfer, affecting IT solution providers, buyers, employees, and contractors alike.
In most jurisdictions, computer programs are considered works protected under copyright law, provided that they meet certain requirements, in a similar way to literary works. The current legal stance, both in Europe and globally, is that to qualify for copyright protection, a given outcome must be a product of human creative activity. This means that if a computer program is generated automatically with minimal or trivial human input, it does not qualify as a protected work under copyright law.
Determining the extent of human creative input in AI-assisted output requires careful analysis, considering the specific AI tool used and the manner and extent of use. The decisive factor is whether a human has a significant enough creative impact on the final output, such as a computer program. This impact is assessed by examining the creative decisions made by humans in the development process and the type and proportion of those decisions in relation to decisions “made” by the machine.
AI tools vary in in the way they operate, from generating complete programs to assisting with minor code completion, quality assessment, optimization recommendations and conceptual brainstorming. In each case, the level of human creative input differs. Generally, even when using AI, programmers make numerous creative decisions, such as accepting, modifying, or rejecting AI-generated solutions and deciding if and when to use the tool, or how to combine different elements. Additionally, the output generated by an AI tool typically reflects prior human input, as humans direct the AI through specific instructions, defining parameters, or selecting from generated options. In other words, in most cases, the final program is the result of a creative combination of many elements (those created by humans and those generated with AI).
Thus, given the current state of technology, in most cases, AI tools are only supportive tools for human work, and it is the human who directs the entire creative process, remaining the main architect and decision-maker.
However, using AI tools typically results in fewer creative elements than when software is created without using such tools. In extreme cases, if the human input becomes so insignificant that the whole is not considered the result of human effort, this outcome will not be subject to copyright. However, this will always depend on a specific case, and each situation must be analyzed individually.
The implications of content created using AI tools not being considered a copyrighted work are extremely significant for all creative industries, especially for the IT industry. If AI-generated software is not recognized as a copyrighted work, copyright transfer or licensing of that element is not possible, which may lead to breaches of contract. Additionally, elements of computer programs not covered by copyright may be freely copied and modified by other individuals, including competitors.
To address these challenges, organizations must adopt proactive legal strategies. This includes legal analysis of AI tool usage in content creation, clearly defining usage policies, and communicating them to staff. Using AI tools prompts additional legal complexities regarding tax benefits: for example, in Poland, tax deductions for software creators are based on the classification of AI-assisted product as copyrightable works (50% tax-deductible costs). Failure to meet this criterion could lead to tax liabilities for programmers and their employers.
Furthermore, the implications of AI use should be considered in business relationships and contracts. Discussing AI tool usage with clients and incorporating AI-related clauses into contracts and internal regulations can help manage risks associated with copyright issues related to AI-assisted content. These measures will mitigate key legal risks and enable safer, more informed use of this revolutionary technology in organizational operations.