Artificial intelligence (AI) is rapidly reshaping how companies operate, creating new opportunities for growth and efficiency but also new risks that investors are beginning to scrutinise more closely.
Whether it’s retail chatbots, fraud detection systems, tools for optimising electricity grids and forecasting energy demand, or software for screening job applicants and supporting productivity, AI is becoming embedded in essential services and even the wider economy.
As adoption accelerates, investors are increasingly assessing not just where companies are using AI, but whether they have the governance, oversight and safeguards to manage the associated risks.
This article is for information only and not personal financial advice. If you’re not sure what’s right for you, a financial adviser can help.
Why AI governance is becoming a stewardship priority
Recent controversies have highlighted what can happen when governance fails to keep pace with change. A US judge recently ruled that Workday must face claims that its AI-powered HR software screened out job applicants in ways that allegedly violated anti-discrimination laws.
Meanwhile, debate about the NHS's use of Palantir's data platform has shown us how AI governance can become a reputational issue, particularly where healthcare data and public trust are involved.
For fund managers, AI is emerging as an increasingly important area of engagement. Some are assessing whether boards have the skills and accountability structures needed to oversee the technology; others are scrutinising its use in higher-risk applications or encouraging greater transparency around governance frameworks.
Learn more about stewardship and engagement on our Responsible Investment hub
We’ve been looking at how fund managers are approaching AI engagement in practice. Although the themes differ, they all reflect a broader shift in investor focus from AI adoption to accountability.
How Fidelity is assessing the governance of AI
Last year, Fidelity International engaged with 31 Australian-listed companies during the AGM season. Recognising that disclosures often provide only a partial picture, Fidelity undertook a dedicated assessment of AI governance practices across companies of different sizes and sectors.
The engagement focused on five areas:
AI strategy
board oversight and capability
risk management
governance and ethical safeguards
workforce impacts
Fidelity used these discussions to benchmark companies against Microsoft's AI maturity framework, assessing how prepared boards were to manage both the opportunities and risks of AI adoption.
The exercise revealed a wide range of preparedness. Although some companies had integrated AI into corporate strategy and developed governance structures to support deployment, many were still in the early stages of building oversight frameworks.
Engagement also identified several recurring concerns, such as unclear accountability, limited discussion of "no-go" use cases, and a reliance on human review processes as the primary risk control.
At the same time, Fidelity also observed examples of emerging good practice – increased board education, investment in AI capability, and stronger links between AI strategy and incentive structures.
Fidelity considers AI governance an increasingly important stewardship issue and intends to continue engaging with boards on oversight, accountability and risk management as adoption accelerates.
How Schroders is engaging with the health insurance sector on responsible AI use
Schroders has been engaging with several large US health insurers following allegations that AI tools were being used in the prior authorisation process.
Prior authorisation is a system through which insurers review and approve certain treatments before they’re provided. Concerns emerged that AI may have influenced approval decisions, potentially affecting patients' access to care.
In response, Schroders engaged directly with insurers to better understand how AI was being used, the controls in place, and the role of human oversight.
The discussions focused on three areas:
whether AI was involved in treatment decisions
the governance and board oversight surrounding these systems
the internal controls designed to prevent inappropriate outcomes
The insurers stated that AI was not used to deny medical treatment requests, although some acknowledged that technology could be used to streamline administrative processes. Engagement also highlighted broader efforts to improve transparency, strengthen oversight and reduce friction in the authorisation process.
Schroders plan to continue engaging on the steps taken to reduce friction in the preauthorisation process, like bundling payments and streamlining data between the insurer and health care providers.
How EdenTree is engaging directly with companies on ethical AI
EdenTree has engaged with Vodafone on the company's approach to ethical AI, the governance structures supporting its use, and the extent to which these practices were reflected in public disclosures.
Following an initial discussion in 2024, EdenTree met with Vodafone a year later to assess progress against its engagement objectives and understand how the company's AI strategy had evolved. Discussions covered the governance in place to manage AI-related risks, employee awareness of responsible AI principles and the company's approach to accountability.
Vodafone reported that it had invested in building AI literacy across the organisation and established a cross-functional Responsible AI programme to ensure shared ownership of AI oversight. The company also highlighted a risk classification framework that applies safeguards based on the specific use case, going beyond minimum legal requirements in identifying higher-risk applications.
Importantly, Vodafone acknowledged that its public disclosure had not kept pace with its internal governance practices and indicated that a formal AI policy was progressing through internal approval processes. For EdenTree, this represented progress towards a key engagement objective – improving transparency around how the company manages AI-related risks and opportunities.