The rapid growth of AI-generated entertainment has raised concerns about the lack of safeguards in place to prevent the creation and distribution of harmful content. In Tanzania, as in many other countries, AI-generated drama series have become increasingly popular, but also repetitive and often featuring storylines that include endless betrayals, abusive husbands, and helpless wives. This has led to questions about who is responsible for ensuring that AI-generated content does not cross the line into harm.
Technology companies have developed usage policies, technical restrictions, and moderation systems to prevent the creation of certain types of content. Social media platforms have their own community guidelines governing what can be published and shared. However, the effectiveness of these systems in preventing the creation and distribution of harmful AI-generated content is uncertain. In Tanzania, as elsewhere, the question remains as to who is watching the AI and ensuring that it does not produce content that is harmful or offensive.
AI guardrails refer to the rules, technical systems, and monitoring processes designed to keep artificial intelligence within certain boundaries. These safeguards can be incorporated at different stages of a model's development and operation, including the data used to train the model, input filtering, and output filtering. Developers can curate and filter training datasets to reduce exposure to harmful or inappropriate material and use additional training techniques, including human feedback, to encourage the model to recognise and respond appropriately to requests that could cause harm.
Despite these safeguards, AI models are probabilistic and can generate responses based on patterns learned from data, and their behaviour can vary depending on the request and context. Developers therefore have to anticipate an enormous range of possible interactions, including those they have never encountered before. This has led to the development of defence-in-depth approaches, which combine multiple safeguards rather than relying on one filter to prevent every possible misuse.
A review of the published usage policies of some of the AI tools people increasingly use in their everyday lives reveals that they prohibit the creation of certain types of content, including child sexual exploitation, graphic violence, and hateful conduct. Runway's usage policy, which was last updated on March 6, 2026, offers a useful case study of the restrictions placed on users of generative video tools. The policy prohibits child sexual exploitation, child abuse, graphic violence, gore, violent extremism, hateful conduct, harassment, self-harm content, and non-consensual intimate imagery.
The International AI Safety Report 2026, a major international review involving more than 100 AI experts and backed by over 30 countries and international organisations, describes this broader approach as defence-in-depth. The report highlights the importance of combining multiple safeguards to prevent the creation and distribution of harmful AI-generated content. However, the report also notes that these systems are not infallible and that developers must anticipate an enormous range of possible interactions, including those they have never encountered before.
As AI-generated entertainment continues to grow in popularity in Tanzania and elsewhere, the question of who is watching the AI and ensuring that it does not produce content that is harmful or offensive remains a pressing concern. While technology companies have developed usage policies and moderation systems to prevent the creation of certain types of content, the effectiveness of these systems in preventing harm is uncertain.
Key points
- AI guardrails are being developed to prevent the creation and distribution of harmful AI-generated content.
- Usage policies and moderation systems are being implemented to prevent the creation of certain types of content.
- The effectiveness of these systems in preventing harm is uncertain.