Australia adopts Anthropic’s Claude Mythos amid uneven cyber safety benefits: systemic disparities in AI-driven security access
Original framing: “Australia now has access to Anthropic’s Claude Mythos. It may improve cyber safety – but not for everyone” — The Conversation - Global
Marginalised users—including refugees, sex workers, and Indigenous activists—are disproportionately targeted by cyber threats yet excluded from the design of safety tools like Claude Mythos. Their lived experiences reveal how 'cyber safety' often translates to increased surveillance and control, particularly for those already policed by state and corporate systems. Grassroots organisations like *Digital Defense Fund* and *TrollBusters* have documented how AI tools fail to protect vulnerable groups, instead amplifying their risks through misaligned threat detection.
The adoption of Anthropic’s Claude Mythos in Australia exemplifies how AI-driven cybersecurity is being repurposed as a tool of neoliberal governance, where 'safety' is redefined as corporate compliance rather than public welfare.
This narrative obscures the historical continuity of colonial cybersecurity paradigms, where Western corporations and states have long treated digital infrastructure as a domain for extraction and control. Indigenous and marginalised communities, who bear the brunt of cyber threats, are systematically excluded from the design and governance of these systems, reinforcing digital divides under the guise of technological progress. The solution lies not in scaling proprietary models but in dismantling their monopoly through publicly owned infrastructure, Indigenous data sovereignty, and participatory design—principles already demonstrated in grassroots movements from Māori *whanaungatanga* networks to Brazilian *favela* cyber collectives. Without these interventions, the AI era will deepen, rather than resolve, the asymmetries of power that define our digital present.