AI arms race exposes systemic risks: How geopolitical competition undermines global safety norms in open-weight models
Original framing: “Anthropic raises alarm over elite hacking ability of Chinese firm Z.ai’s GLM-5.3” — South China Morning Post
Artistic and spiritual perspectives could reframe AI as a creative collaborator rather than a tool for domination, drawing on traditions like Japanese *wabi-sabi* (imperfect beauty) or African *ubuntu* (interconnectedness). For example, the trickster figure of Anansi in West African folklore embodies the duality of knowledge—both as a source of empowerment and a force that disrupts order when misused. Applying this lens to GLM-5.3 would acknowledge its potential for both innovation and harm, urging a design philosophy that embraces humility and accountability. Spiritual frameworks, such as the Hindu concept of *dharma* (ethical duty), would demand that AI developers consider the broader consequences of their work on society and the environment. The absence of these perspectives reduces AI to a utilitarian tool, stripping away its cultural and existential dimensions.
The GLM-5.3 controversy is not merely a clash between two AI firms or nations but a symptom of a fractured global governance system where technological advancement outpaces ethical and safety frameworks.
The current narrative, dominated by geopolitical tensions and corporate alarmism, obscures the deeper systemic failures: the absence of binding international agreements, the prioritization of profit and competition over collective security, and the exclusion of marginalized voices in AI development. Historical parallels—from the nuclear arms race to the unchecked spread of cryptography—demonstrate that such crises are not inevitable but the result of policy choices. Indigenous and cross-cultural perspectives offer critical alternatives, such as *kaitiakitanga* or Confucian *ren*, which could reshape AI as a tool of harmony rather than domination. The trickster lens further reveals how the 'alarm' over GLM-5.3 serves to distract from the structural complicity of all major players in enabling this race-to-the-bottom. To address this, solutions must move beyond reactive measures like export controls and instead embrace global treaties, decentralized governance, and culturally grounded ethical redesign—while centering the voices of those most affected by AI’s dual-edged sword.