Report
Document cover

Humanising the algorithm: the safety implications of algorithmic decision-making on delivery riders

Project report – Phase 1 and 2: review of academic research literature, stakeholder interviews
Publisher
Gig economy Working conditions Occupational health and safety Risk assessment Algorithmic regulation Road safety Autonomous technologies Victoria
Resources
Attachment Size
download linkHumanising the algorithm 1.12 MB
Description

This report presents independent research on the safety, governance and public impacts of algorithmic management in app-based food-delivery work. It draws on a comprehensive analysis of global academic research, and on interviews with key stakeholder organisations with an interest in road safety in Victoria, including the Department of Transport and Planning, Bicycle Network Australia, Swinburne University of Technology, TAC, WorkSafe Victoria, Victoria Police and the councils of Yarra, Stonnington and Melbourne, conducted between August and September 2025. 

The report provides empirical insights into how platform algorithms shape rider behaviour, safety outcomes and interactions with consumers and the wider community. The study highlights an ecosystem of road safety risks that are compounded by a lack of clear ownership: no single actor, platforms, regulators, urban planners, food suppliers, holds responsibility for ensuring safe delivery conditions and that is problematic. 

The findings offer policymakers, regulators, industry stakeholders and platform operators’ actionable evidence to support fairer, safer and more accountable gig-delivery ecosystems. The report proposes further research to develop a world-first, safety-focused algorithmic audit framework. 

Recommendations

  1. Establish a Stakeholder Advisory Committee.
  2. Develop independent algorithmic audit approaches.
  3. Strengthen advocacy for governance standards and policy reform.
  4. Take immediate and coordinated action for rider safety and infrastructure improvement in high-density areas.
Publication Details
License type:
CC BY
Access Rights Type:
open