Generative artificial intelligence (AI) is creating new legal headaches for brands in Ghana. Companies that use AI in their marketing face challenges related to copyrighted material. AI systems can also generate content that copies or imitates famous people or registered brand names. This is not a future problem; it is happening now and affects businesses across Africa.
AI technology is increasingly integrated into how businesses create advertising. This brings up complex issues concerning intellectual property, who is responsible when things go wrong, and how to regulate this technology. This article looks at the legal problems that arise when AI software is built using data that includes copyrighted works. It also examines AI used in advertising campaigns.
The ownership of content created by AI is uncertain. There are also risks of violating existing copyrights. Accountability frameworks for AI-generated content and advertising are being developed. Evolving court decisions show that AI-generated content may not get copyright protection. However, AI creators could still be responsible if their outputs copy someone else's work.
Experts are considering how to assign blame. This includes AI developers, the platforms that host AI, advertisers, brands, and even consumers. Scenarios involving AI that spreads false information or creates fake images, known as deepfakes, are a growing concern. Regulators worldwide are looking at ways to make AI developers and brands more accountable. They also want more transparency in how AI is used.
A key issue is AI training data. Generative AI models learn from vast amounts of data. This data is often collected from the internet. Sometimes, this data includes copyrighted material without permission. The legal question is whether using this material to train AI counts as fair use or copyright infringement.
In 2025, a United States court ruled in Bartz v. Anthropic. It said that legally sourced materials could be used for AI training under fair use. However, using pirated books was found to be copyright infringement. Anthropic had to destroy pirated datasets as part of a $1.5 billion settlement. This was the largest settlement for AI copyright issues.
Another US court noted that Meta's use of copyrighted material might be transformative. However, the court stressed that companies must pay copyright holders if this use is essential. In the United Kingdom, Getty Images sued Stability AI. Getty Images largely lost its main copyright claims. It did win a narrow ruling on trademark infringement. This UK decision offers some legal protection to AI developers regarding training data. It also highlights that AI companies must not reproduce protected watermarks in their AI-generated outputs.
In Nigeria, the Advertising Regulatory Council (ARCON) has warned digital marketers. This is because of fake AI-generated advertisements on social media. These ads often make misleading health claims, which pose a significant public health risk.
Brands in Ghana should be aware that their copyrighted materials might be used to train AI models without them knowing. It is recommended to check training datasets regularly. Brands should also actively protect their intellectual property rights. This can be done by sending take-down requests when necessary.
Regarding AI-generated outputs, the question of ownership is critical. On March 2, 2026, the United States Supreme Court chose not to review Thaler v. Perlmutter. This decision confirmed that content created solely by AI, without significant human creativity, cannot be copyrighted. Therefore, logos or advertising copy generated entirely by AI for a brand will not receive copyright protection.
Despite this, US courts are becoming more open to claims of output-based infringement. These claims argue that an AI's output directly violates existing copyrights. This distinction is important. While AI creators might not own their creations, they can be held responsible if their AI systems reproduce copyrighted material belonging to others.
Ghanaian brands are advised to document any human creative input in AI-assisted work. Relying only on AI for core brand assets is legally risky. Trademark law focuses on commercial use. In the context of AI, liability can occur if trademarked materials are used in training data. It can also happen if AI-generated content includes protected marks. Furthermore, if an AI product is given a name that conflicts with an existing trademark, that can lead to legal issues.
The case of Cameo v. OpenAI illustrates these trademark risks. Cameo, a platform for personalised celebrity videos, sued OpenAI. This happened after OpenAI introduced a feature in its Sora video app. This feature allowed users to create AI videos featuring celebrity likenesses. The court's decision in such cases will shape how brands navigate AI's impact on their identity and legal standing.