Ed Newton-Rex, a prominent figure in the global artificial intelligence sector and a former researcher at Google DeepMind, has issued a stark warning regarding the existential threats posed by rapidly advancing AI systems. Newton-Rex suggests that the window for humanity to implement effective safeguards is closing as the pace of development outstrips the ability of governments to regulate the technology.
The warning comes at a time when the commercial race for Artificial General Intelligence (AGI) has reached a fever pitch, with billions of dollars in venture capital flowing into labs that are prioritising speed and scale over safety protocols. Newton-Rex, who previously served as the Vice President of Audio at Stability AI before resigning over ethical disagreements, argues that the trajectory of current models could eventually lead to outcomes that are catastrophic for human civilization.
This sentiment echoes concerns raised by other industry pioneers, including Geoffrey Hinton and Yoshua Bengio, often referred to as the “godfathers” of AI. However, Newton-Rex’s position is particularly notable due to his recent frontline experience in both big tech and the generative AI startup ecosystem. He has increasingly focused on the intersection of intellectual property and safety, founding Fairly Trained, a non-profit that certifies AI models that do not rely on copyrighted data without consent.
The core of the current debate hinges on the “alignment problem”—the technical challenge of ensuring that an AI’s goals remain perfectly synchronised with human values. As models become more autonomous and capable of reasoning, the risk of unintended consequences increases. For business leaders and investors, this is no longer a theoretical concern; it is becoming a matter of long-term operational viability and regulatory compliance.
Global Regulators Grapple with AI Governance Frameworks
Governments across the world are attempting to catch up with the technological curve. The European Union’s AI Act represents the first comprehensive attempt to categorise and regulate AI based on risk levels. However, critics argue that such legislative frameworks are often too slow to adapt to the breakthroughs occurring in private laboratories every few months. While the EU focuses on data privacy and transparency, the “existential risk” or “x-risk” advocates believe the focus should be on the potential for autonomous systems to bypass human control entirely.
In Africa, the conversation is shifting toward how these global risks and regulations will impact local digital economies. Nigeria recently took a significant step by launching its National Artificial Intelligence Strategy, aiming to position the country as a leader in responsible AI adoption. The challenge for African policymakers is twofold: ensuring that the continent is not used as a testing ground for unsafe AI while simultaneously ensuring that stringent global regulations do not stifle local innovation or access to advanced tools.
The financial implications of AI safety are also manifesting in the markets. Major tech firms are now required to include AI-related risks in their annual reports. Companies like Microsoft, Google, and Meta face growing pressure from shareholders to disclose the safety testing protocols used for their frontier models. This pressure has led to the formation of the AI Safety Institute in the UK and similar bodies in the US, which aim to provide independent evaluations of high-risk systems before they are released to the public.
Newton-Rex’s resignation from Stability AI in 2024 was a watershed moment for the industry’s ethical discourse. He argued that the use of creative work without a licence to train models was not only a violation of copyright but also an indication of a broader disregard for human agency within the AI development pipeline. He contends that if companies cannot respect basic intellectual property rights, they are unlikely to exercise the necessary caution when it comes to more abstract existential risks.
The debate is currently split between “accelerationists,” who believe that AI will solve humanity’s greatest challenges and should be developed as quickly as possible, and “doomers” or safety advocates who urge a more cautious approach. For the business elite, the middle ground involves “Responsible AI”—a framework that balances profit and innovation with rigorous ethical auditing.
As the capabilities of Large Language Models (LLMs) continue to expand into multi-modal reasoning and autonomous agentic behaviour, the calls for a global, binding treaty on AI safety are growing louder. The next stage of this development will likely involve the implementation of “compute governance,” where the massive hardware resources required to train advanced AI are monitored to prevent the creation of unaligned or dangerous systems. For now, the warnings from researchers like Newton-Rex serve as a reminder that the window for meaningful human intervention is narrow.
Explore more Tech stories and analysis from Business Elites Africa.
