Download the full submission here.

RIAA thanks the Australian Government for the opportunity to comment on the future of artificial intelligence (AI) training and large data centres (the consultation). RIAA agrees that a nationally consistent approach to legislation and standards is necessary to provide the certainty required for investment attraction and to minimise investment risk. RIAA welcomes the Prime Minister’s announcement of national AI laws in Australia’s interests.

AI and its supporting infrastructure can create material financial, environmental, social and system-level risks for institutional investors and their beneficiaries. Clear governance, credible safeguards and decision-useful disclosure are therefore important to enable investors to assess these risks and support durable, responsible innovation.

RIAA recommends that the Australian Government:

• establish nationally consistent, enforceable requirements for large data centres and frontier AI development, with obligations proportionate to material risks and impacts;
• ensure new data centre demand is supported by additional renewable energy, appropriate storage and flexibility, while avoiding the transfer of attributable costs to households and other businesses;
• require transparent and comparable reporting on material energy, water, emissions, environmental and community impacts;
• require early and meaningful engagement with affected communities and rights holders, including the protection of First Nations Peoples’ rights, Country and cultural heritage; and
• establish risk-based, human-rights-centred safeguards across the AI lifecycle, with clear accountability, due diligence, testing and assurance, monitoring, transparency and access to remedy.

A just and equitable transition should underpin these measures. Managing impacts on workers, communities, consumers, First Nations Peoples and the environment will be important to maintaining social licence and ensuring that the benefits of AI are durable and broadly shared.

General comments

Risks of AI to institutional investors

The deployment of artificial intelligence creates material risks for investors because inadequate design, governance or use can affect company value, portfolio resilience and the stability of the wider economy. RIAA’s Artificial Intelligence and Human Rights Investor Toolkit, released in May 2024, highlights that AI-related harm is not only an ethical concern. It can translate into legal liability, regulatory intervention, operational disruption, remediation costs, loss of customer trust and damage to social licence. These risks can arise across the AI value chain—including developers, providers of data and infrastructure, companies deploying AI, and businesses whose products or services are reshaped by it—so exposure may be broader and less visible than investors expect.

At a company level, investors need to consider whether AI is being deployed with effective Board oversight, human rights due diligence, testing, transparency and accessible remedy. Poorly governed systems can produce biased or discriminatory outcomes, invade privacy, enable surveillance, amplify misinformation, undermine worker rights, misuse intellectual property or make consequential decisions that affected people cannot understand or contest. Cybersecurity failures, unsafe outputs, unreliable models and AI “hallucinations” can also compromise services and decision-making. Where companies move quickly without adequate capability, controls or human oversight, expected productivity gains may be offset by implementation failure, litigation, regulatory penalties, reputational damage and stranded technology investment.

AI also creates significant environmental, social and system-level risks that may affect diversified portfolios. Training and operating advanced models can drive substantial demand for electricity, water, land, minerals and data centre infrastructure, increasing exposure to emissions, resource constraints, community opposition, impacts on nature and Country, impacts on cultural heritage and higher costs for users. Rapid automation may disrupt employment and wages, deepen inequality and weaken the consumer demand on which many business models depend. At the same time, AI-generated content and misinformation can erode trust in institutions and markets. Concentration of compute, data and model ownership can entrench market power and create dependencies on a small number of providers, while interconnected use of similar models may produce correlated errors and new sources of systemic risk.

Developments such as growing inequality, increased unemployment and undermining of market confidence create additional and compounding uncertainties and risks for investors, particularly universal owners. Given the rapid uptake and deployment of AI across our society, these risks cannot be divested away.

For investors, the central issue is therefore whether investee companies can demonstrate responsible deployment across the full AI lifecycle. Decision-useful disclosure should explain accountability, material use cases, affected stakeholders, testing and assurance, data and cybersecurity controls, serious incidents, environmental impacts and remediation. Investors can use stewardship, voting, engagement and capital allocation to press for safeguards proportionate to the severity, scale, likelihood and remediability of harm. Without clear governance and credible evidence that controls work in practice, investors may be unable to price AI-related risks accurately or distinguish durable innovation from short-term claims, leaving beneficiaries exposed to avoidable financial and systemic harm.

Strong, enforced baseline laws regulating the expansion, deployment and use of AI will go some way to mitigating the significant reputational, operational, regulatory, financial and social risks facing investors investing in the AI value chain. This current consultation should translate high-level government principles into legislated national standards.

Risks of losing social licence

AI development and deployment depend on public confidence that benefits will be widely shared and that harms will be prevented, addressed and remedied. Social licence may be lost where communities, workers and consumers experience AI as something done to them rather than developed with them. This risk is pronounced where deployment is opaque, accountability is unclear or people cannot challenge consequential decisions. The Australian Government’s own approach to responsible use of AI recognises transparency, governance, risk assurance and public trust as mutually reinforcing requirements.

For investors, loss of trust can become a material investment risk. Lack of trust can drive community opposition, workforce resistance, customer disengagement, regulatory intervention, project delay and higher compliance or remediation costs. It may also constrain adoption and weaken the durability of anticipated productivity gains. The risk is heightened where AI-related infrastructure imposes local costs (such as pressure on energy, water, land, nature or cultural heritage) without credible participation, benefit-sharing or safeguards.

Examples: The impact of losing social licence

• Australia: A Perth proposal was abandoned after nearly 1,900 public objections and a recommendation for refusal.
• United States: A US$1.5 billion proposal in San Marcos, Texas, was halted following strong opposition concerning water scarcity, grid impacts and environmental sustainability.
• Ireland: Electricity-system pressures prompted restrictions on new data centre connections; one delayed Dublin project reportedly exposed its developer to a €152 million compensation claim.
• Europe: Over 70 projects were reportedly rejected or restricted in early 2026, while opposition affected about US$42 billion of investment through delays and cancellations.

A just and equitable transition should therefore be treated as a core condition of responsible AI adoption, not as a response after disruption occurs. The OECD AI Principles call for inclusive growth, sustainable development and wellbeing, human rights, transparency, safety and accountability, and recommend building human capacity and preparing for labour-market transition. The International Labour Organization similarly emphasises proactive policy, job quality, fair transitions, social dialogue and effective regulation; its research also indicates that impacts may be unevenly distributed, including greater exposure among women because of occupational patterns.

Government and companies should assess distributional impacts before deployment; engage affected workers, unions, communities, First Nations Peoples and other rights holders early and meaningfully; provide accessible information and genuine influence over material decisions; invest in skills, reskilling and pathways to good work; protect job quality, privacy, equality and human oversight; and establish effective grievance, contestability and remedy mechanisms. The Australian Human Rights Commission has likewise called for AI innovation to be consultative, inclusive and accountable, supported by robust human rights safeguards.

For investors, credible transition planning includes Board accountability, human rights due diligence, workforce and community impact assessment, measurable commitments, disclosure of outcomes and evidence that affected groups have participated in decision-making. The UN Guiding Principles on Business and Human Rights provide an appropriate baseline through the duties to protect, respect and remedy. These measures can help preserve social licence, reduce implementation and regulatory risk, and ensure that AI-derived value is created in a way that is durable, legitimate and fairly shared.

‘Download submission’ for responses to consultation questions.

‍

‍