As artificial intelligence moves from answering questions to taking action, the safety debate is shifting from chatbots to autonomous agents
By Unity Times Technology Desk
Grow your business with us
Artificial intelligence is entering a new phase. The question is no longer simply whether machines can answer questions. It is increasingly whether they can act on behalf of humans — and whether humans can safely control what happens when they do.
That is why recent developments around OpenAI deserve attention beyond Silicon Valley.
The company has been expanding protections for teenage users while simultaneously confronting emerging questions around increasingly capable AI systems, including the risks associated with autonomous or agent-like behaviour.
OpenAI says accounts identified as belonging to users under 18 now receive additional protections, including safeguards around graphic violence, self-harm, dangerous challenges, unhealthy body-image content and sexual or romantic role-play.
Parents can also link accounts and establish controls such as quiet hours, turning off voice mode, disabling memory and limiting image generation.
For Africa, where the continent has one of the world’s youngest populations, this debate matters enormously.
Why Africa should care
Nigeria is already one of Africa’s largest technology and internet markets. Millions of young Nigerians use AI for schoolwork, coding, writing, business, research and creative work.
That creates enormous opportunities. A student who cannot afford private tutoring can use AI to understand mathematics. A young entrepreneur can use it to draft a business plan. A farmer can access information. A developer can build products faster. But access also creates risks.
Young users may encounter inappropriate material, become over-dependent on AI, accept inaccurate answers as fact or share sensitive information without understanding the consequences.
OpenAI says nearly nine in ten teens using ChatGPT use it for learning, information, skill-building or productivity in a given week. That statistic illustrates the scale of the transformation. AI is no longer an occasional technology for young people. It is becoming part of their education and everyday life.
Then comes the bigger question: AI agents
The safety challenge becomes more complicated when AI systems move beyond conversation.
An AI agent can potentially interact with websites, software, files and other digital environments. That creates opportunities for productivity — but also new risks.
OpenAI recently disclosed incidents identified during independent cyber evaluations in which particular testing configurations and reduced safeguards allowed models to operate beyond intended testing boundaries, including accessing the public internet. The company stressed that the configurations did not represent ordinary public deployment.
The significance is not that AI has suddenly become uncontrollable. It is that testing powerful systems is becoming a safety discipline of its own. The more capable AI becomes, the more important it is to know what happens when systems encounter unexpected situations.
AI and the African development question
For Africa, the debate should not be reduced to whether AI is dangerous. The more useful question is: How can Africa benefit from AI without importing its risks blindly?
African countries need:
- stronger digital literacy;
- AI education in schools and universities;
- local-language AI systems;
- privacy protections;
- cybersecurity capacity;
- clear rules for children and teenagers;
- responsible use of AI in government;
- and safeguards against automated discrimination.
Nigeria cannot afford to be merely a consumer of AI technology. The country needs to become a builder, regulator and informed participant in the AI economy.
The coming battle is trust
The future of AI will ultimately depend on trust. Can parents trust AI around their children? Can students trust AI answers? Can governments trust automated systems? Can businesses allow AI agents to perform sensitive tasks? Can users trust companies to protect their private information?
Those questions are becoming more important as AI becomes more deeply embedded in everyday life.
OpenAI’s teen-safety measures — including age prediction, parental controls and stronger safeguards — represent one attempt to answer some of them.
But technology companies cannot solve the problem alone. Governments, schools, parents, developers and users all have roles to play.
For Africa, the message is simple: The AI revolution is already here. The question is whether we will enter it prepared — or merely arrive after the rules have been written by others.
About The Author