Why AI Safety is a Global Priority
Tech News - Voices calling for global collaborative regulatory control over AI: “I have major issues with the people and the companies behind AI,” says Ioannis Ioannou, associate professor of strategy and entrepreneurship at London Business School. “It's a huge concentration of power.” Ekaterina Abramova, adjunct assistant professor at London Business School, is particularly concerned about AI making consequential decisions in the military. “Do we really want to trust agentic systems with lethal autonomous weapons?” she asks. Abramova points to an exemption in the EU AI Act for national security, which she sees as a significant gap. “We have a government willing to regulate businesses but not necessarily willing to regulate itself,” Abramova says. Here is why AI Safety is a Global Priority, according to the report ¨Solving Europe's AI adoption puzzle¨ by Sifted:
Do we really want to trust agentic systems with lethal autonomous weapons?
EU member states are already putting AI to work in national security. Germany, for example, has said its intelligence service will use AI to collect, store and analyse more information, and to retain it for longer. Ekaterina Abramova, of the London Business School argues that governments need to do more to oversee how AI is developed and deployed. She would like to see a UK AI Act, alongside greater collaboration between governments and leading AI researchers. Policymakers, she says, need to understand the potential problems with the most advanced models directly from the scientists developing them — so they can anticipate what will need regulating rather than react once it has gone wrong.
The next step, in her view, would be global treaties governing how governments use AI. China and the US would ideally be the keystones of any such agreement, although Abramova acknowledges the difficulties: “Historically, we've not been great at global collaboration.” Nevertheless, she argues, “it's naive to think we will control artificial superintelligence unless the priority changes from winning the AI race, to getting it right.” She also calls for a global scientific body modelled on the Intergovernmental Panel on Climate Change (IPCC), bringing together scientists from around the world to assess AI's risks. A version of this already exists in the International AI Safety Report, commissioned by the UK government and chaired by Turing award-winning AI scientist Yoshua Bengio. Another possibility is mandatory reporting of serious AI incidents, allowing countries to share information and learn from one another. Ultimately, she says, this could evolve into a verification and inspection treaty under which representatives from one country could inspect another's AI development. All of this would inevitably create more red tape for businesses — an unavoidable tradeoff:
“Compliance is a real cost and it would bite small and medium firms the hardest, but better safe than sorry. We need to regulate more and not wait until something bad happens to make decisions.” - Ekaterina Abramova, adjunct assistant professor at London Business School.
