<p>The incorporation of Artificial Intelligence (AI) into welfare regimes holds the promise for notable developments in the effectiveness of social service delivery. Meanwhile, it also brings the risk of inequality and encouraging distrust. This research analyses the algorithmic welfare state, as the strengths of AI in predictive analytics, data collection, model-based representation, and computerized decision-making redefine governmental operations as compared to the digital welfare state involving a wider range of digital technologies. Analysing AI for each context namely developing, universal, and means-tested welfare regimes, the research reveals the capability for enhancing both scaling and accuracy and, at the same time, revealing the risk of algorithmic bias, digital invisibility, and eroding trust due to non-transparent systems. These findings contravene the welfare values of fairness, trust, and agential freedom, highlighting the need for ethical governing. The research introduces a conceptual framework accentuating the need for transparency, distributive equity, and human oversight to reduce potential peril, spotting opportunities for further research and clarifying the boundaries, thus adding value to the discourse around technology-facilitated governing and social justice.</p>