High stakes, low guardrails: reporting on AI in a regulatory vacuum

AI is changing how companies communicate with stakeholders. As reporters, we're all talking about how we can leverage AI to optimise our processes. How we can augment the outcomes for audiences, who, in turn, are increasingly using AI to read reports or, let’s face it, skip straight to juicy content. But are we giving enough consideration to how we talk about AI as an actual topic within Annual, Integrated and Sustainability reporting? 

 

AI is a strategic imperative, a test of governance and risk management and a pertinent ESG topic that needs to be high on the agenda for all corporate communications and reporting teams.

 

No doubt, it needs to be addressed in reporting. But formal standards and established metrics on this topic are nascent and frankly lacking.

 

In this apparent vacuum, companies are taking it upon themselves to test out content and communication styles. There are advantages to this moment of experimentation: by staking a claim early and being forthcoming about how AI is transforming business, reporters can engage audiences in memorable and unique ways. But without the rigour of established regulation, we’re also already starting to see bad habits take shape. How can companies be strategic and considered when it comes to reporting on AI to maximise impact and preparate for looming regulation?

 

A Swiss cheese of reporting standards?

 

We’re used to regulation offering a level of guidance but the regulation around AI and reporting is still highly nascent, with a notable absence of established metrics and formal standards.

 

There has been a degree of broad guidance on AI communications. First from the EU AI Act, which came into effect from August 2024. The Act provides the first international regulatory framework for AI and areas such as the Transparency Obligations in Article 50 or the Notification Procedure in Article 30 relate somewhat to communications. Similarly, the OECD AI Due Diligence Guidance released in February 2026 offers light-touch instruction under ‘Step 5: Communicate actions to address impacts’. Crucially however, neither of these frameworks focus on reporting and therefore fail to provide a targeted, explicit basis for corporate reporting on AI.

 

Reviews of existing sustainability standards and requirements signal a critical ‘reporting gap’ around AI. A 2024 report from GRI evaluated its own standards as well as the wider regulatory landscape, including ESRS, SASB and CDP, and found that “the GRI Standards and other internationally recognised sustainability reporting frameworks are lagging behind in setting reporting expectations for the adoption of digital technologies and the implications of the transition”.

 

This patchwork of partial coverage isn’t enough to accurately reflect the impacts, risks and opportunities associated with AI.

 

Trends and trip-ups in early phase reporting

 

Regulation will swiftly catch up and standards will evolve. But for now, this relative freedom means companies are operating in an experimental phase, testing the most effective ways to measure and report on AI. So, what changes has this driven in both annual and sustainability reporting? Best practices are emerging, but so are early pitfalls. Both can help form a blueprint market standard.

 

What we’re seeing in Annual and Integrated reporting.

 

A 2025 publication “Decoding AI Disclosure” evaluates AI reporting from the largest 50 European companies. Early year-over-year comparison indicates AI disclosure is growing in multiple ways. In 2025 companies were found to be providing deeper insights: significantly more detailed information and broader coverage, reporting AI across a wider span of reporting sections. While more good quality disclosure is rarely a bad move, reporting should always be concise and considered. Technology is iterating at lightning speed and reckless or superficial reporting could put a company in a sensitive and exposing position.

 

Every company in the research cohort was found to be reporting on AI through a strategic lens and a significant majority (45 out of 50) report on how AI is being used. While investors want to know how AI is being adopted to advance strategic opportunities, they also want reassurance investment is being tracked against ROI metrics. According to Deloitte’s 2025 survey, 91% of companies plan to increase investment in AI this year, yet most respondents indicated ROI was lagging behind typical payback expectancy. Companies need to be mindful of jumping on the AI hype train without reporting specifically on the anticipated or achieved returns.

Coverage of key areas of governance has also increased significantly over the past two years: reporting on policy, Board oversight, knowledge development and risk management all increased by about 50%. In fact, the latest FRC Annual Review of Corporate Governance Reporting found that 86% of companies include AI within governance. Interestingly, only 20% of companies identified AI as an emerging risk, while 3% elevated the topic to a Principal Risk. Just five companies of 50 reported on AI audits. Policymakers and regulators are watching closely and want to see governance measures are in place to protect against risk and negative impact; customers and suppliers want confirmation data is being handled securely and ethically, and their privacy is being protected. Are companies taking AI risks seriously enough?

 

The trends shaping sustainability reporting

 

A 2025 Teneo report into the state of sustainability reporting showed that more companies are including disclosure around the responsible use of AI in sustainability reports. Disclosure increased by more than 50% compared with the previous year. As it stands, AI is frequently confined to a governance matter, but as the environmental and social impacts become more widely known, it’s important that it stretches across sustainability reporting. Employees and recruits will be understandably concerned around job security and therefore want to know what skills are being promoted, valued and boosted. They want to see leadership taking accountability, spearheading implementation with sincere consideration for the social and environmental impacts.

 

Being too slow, too silent or too siloed around AI could well be perceived as ‘greenhush’. But at the other end of the spectrum, some companies are being blindly evangelical about the opportunities AI brings to sustainability, overstating the benefits and ignoring the costs. A recent Guardian article calls out big tech players for being dangerous perpetrators of this kind of greenwash. AI has significant negative impacts and risks when it comes to matters such as carbon emissions, water usage, e-waste, biodiversity, human rights, cybersecurity and data ethics: reporting should reflect this.

 

The solution: structured experimentation

 

While regulation slowly takes shape, and market standards evolve, we advocate for a bold and balanced approach to AI reporting that blends experimentation with creative communications with structured, strategic thinking.

To help, we’ve compiled three starting principles for AI in reporting.

 

Balance breadth with depth.  

Superficial disclosure will leave audiences under-served and won’t be nearly as effective as delving into selective detail and offering a nuanced narrative. Don’t proclaim exclusively on AI in strategy and use cases without speaking to the governance measures and risk mitigation you’ve put into place.

 

In Zurich’s 2025 Annual Report, the company includes an editorial spotlight feature early on to showcase how it is using AI as a catalyst for strategy development and delivery. This is proffered alongside a similar spotlight on cyber resilience, showing a clear balance between opportunity and risk.

 

Don’t reinvent the wheel

In the absence of formal reporting regulation, approaches could be adapted from tried and tested formulations. An organisation could treat AI as a material topic, or as a principle or emerging risk, and report accordingly. Reporters could choose to structure disclosure by the TCFD / TNFD four-pillar approach: Governance, Strategy, Risk Management, and Metrics & Targets. Equally the GRI environmental and economic Topic Standards provide a strong starting basis for sustainability reporting.

 

Regardless of the formulation you adopt however, be wise to where the gaping regulatory holes are and start to fill these where you can. Regulation will catch up at some point.

 

Avoid the hype

Investors don’t want to see companies blindly spending. Work out where the return is expected and plan for how to measure and improve on it. Likewise, be transparent over the risks and the negative impacts across the ESG spectrum.

 

ASML’s 2025 Annual Report takes a balanced stance on the sustainability impacts of AI, acknowledging the fresh challenges that come with it. AI is also a notable key feature of the CFO statement, with clear indication that ASML is monitoring ROI.

 

Looking to take advantage of this moment? This phase of regulatory freedom won’t last long, your audiences are interested and your peers are already experimenting. Reach out to RY for support delivering bold and balanced AI narratives in Annual, Integrated and Sustainability Reporting.