Ghana's Copyright Act, 2005 (Act 690) in the Age of Artificial Intelligence: Text and Data Mining, Fair Dealing, and the Case for Legislative Reform

The emergence of large language models (LLMs) and other generative AI technologies has thrust text and data mining (TDM) into the centre of copyright law globally. In Ghana, the Copyright Act, 2005 (Act 690) contains no provision explicitly addressing TDM or automated machine learning processes, leaving rights holders without clarity and AI developers without certainty. This paper examines two interlocking doctrinal questions: first, whether the permitted-use and fairpractise/dealing provisions of Act 690 can be reasonably interpreted to accommodate TDM and AI training; and second, what a Berne-compliant, Ghana-appropriate TDM exception would look like in statutory language and in practice. Situating these questions within the broader discourse on rights protection in the data economy, the paper draws selectively on the EU's DSM Directive and comparable instruments primarily to frame policy options, before dedicating its principal analysis to how Act 690 may be reformed. It argues that legislative silence exposes Ghana to the dual risks of innovation suppression and unchecked creative-industry exploitation, both of which undermine the country's digital economy ambitions. The paper proposes a targeted statutory amendment, a collective-licensing framework, opt-out mechanisms, and institutional capacity measures, including engagement with ARIPO and the AfCFTA IP Protocol, as a coherent, Bernecompliant reform package calibrated to Ghanaian socioeconomic realities.