Desk Report
Oniket Research Group
Artificial intelligence is no longer a distant possibility for Bangladesh’s creative and digital economy. It is already here, producing text, images, music, and software with astonishing speed and diminishing human involvement. This unprecedented capability poses a fundamental legal challenge: can Bangladesh’s copyright framework, designed around the premise of human creativity, adapt to a reality where machines generate expressive works?
The Roots of the Problem
Copyright law across the world rests on a philosophical assumption that creative works emerge from human intellectual effort. The Berne Convention and the TRIPS Agreement both implicitly assume human authorship. Two major philosophical traditions underscore this: John Locke’s labour theory, which holds that individuals earn property rights through personal effort and skill, and Immanuel Kant’s personality theory, which treats creative works as extensions of the author’s identity. When a machine produces a painting, writes a poem, or composes a melody, neither framework comfortably applies. The law simply has no established mechanism to attribute authorship to a nonhuman entity, nor does it have clear grounds to assign ownership to someone whose contribution was limited to typing a prompt.
Bangladesh’s Copyright Act of 2023 modernised several aspects of digital copyright protection, yet it remains conspicuously silent on AI generated content. The Act continues to presuppose a human author behind every protected work. This silence is no longer tenable. Generative AI tools are already widely accessible in Bangladesh, and they are being used to produce commercial content at scale. Without clear rules, disputes over ownership and infringement will multiply, and courts will face ambiguities the statute never anticipated.
The Nature of the Threat
The challenge extends well beyond simple copying. AI systems can now imitate style, tone, structure, and expressive identity with a fidelity previously associated only with human creators. When an AI model replicates an artist’s visual style or a writer’s narrative voice, it erodes the expressive identity that copyright seeks to protect. Moreover, investigations have revealed that AI models can generate highly realistic reproductions of sensitive documents from minimal prompts, raising urgent concerns about data privacy alongside intellectual property. In a legal framework still anchored to human authorship, such capabilities create structural instability. If expressive works can be produced with negligible human intellectual contribution, the core question becomes unavoidable: what exactly is copyright law protecting?
Reforms Bangladesh Should Pursue
First, Bangladesh should introduce a mandatory disclosure requirement for AI generated or AI assisted works. Any creator, publisher, or applicant for copyright registration who has substantially used generative AI tools in producing a work must declare this at the point of registration or commercial publication. Noncompliance should carry meaningful penalties, including the possibility of invalidated copyright claims. Disclosure does not resolve every question, but it establishes transparency, which is the essential foundation for any regulatory effort.
Second, the law should adopt a dominance test for human intellectual contribution. Rather than banning AI assistance outright, the statute should stipulate that copyright protection is available only when meaningful human creativity, selection, arrangement, editing, or judgment remains the dominant force behind the final work. Courts and adjudicating bodies should evaluate this on a case-by-case basis, examining factors such as the degree of human creative control, the extent of editorial refinement, and whether the output reflects identifiable human intellectual decisions. This approach preserves flexibility as AI technology evolves and avoids the impossible task of drawing a rigid line between human and machine production.
Third, Bangladesh should establish a sui generis framework for works that are generated autonomously by AI without meaningful human intellectual contribution. Traditional copyright protection should not apply to such outputs, because extending full copyright to machine generated production would gradually weaken the philosophical foundations upon which the entire copyright system depends. Instead, a separate legal category, with shorter protection durations and narrower exclusive rights, could provide limited commercial incentives without conflating machine output with human creativity. This sui generis regime would also define permitted uses more narrowly, preventing the monopolisation of AI generated content in ways that could stifle access and innovation.
Fourth, the government should create a standing regulatory body or expert committee specifically tasked with monitoring AI’s impact on intellectual property. This entity would review emerging technologies, consult stakeholders, and recommend periodic amendments to keep pace with rapid technological change. The committee should include technologists, legal scholars, creative industry representatives, and civil society voices to ensure that policy reflects diverse interests rather than narrow commercial priorities.
Fifth, Bangladesh should invest in public awareness and judicial training. Judges, lawyers, and creators must understand how generative AI functions, what can and cannot do, and why existing legal concepts such as originality and authorship require recalibration. Without this knowledge base, even well-crafted statutes will suffer from inconsistent interpretation and enforcement.
The Path Forward
The intersection of artificial intelligence and copyright law is not a problem that will resolve itself through inaction. Bangladesh has a narrow window to develop a balanced, human centred legal framework before AI generated content saturates its creative markets and its courts. The reforms outlined above, mandatory disclosure, a dominance test for human contribution, sui generis protection for autonomous AI output, a dedicated monitoring body, and sustained education, offer a practical path. They respect the traditions of copyright while acknowledging that those traditions alone cannot govern a future where machines create. The choice is straightforward: act now with deliberate, calibrated reform, or allow ambiguity to erode the very concept of creative ownership.
