AI can enhance XBRL reporting – but it can’t replace it
The idea that AI doesn’t need XBRL is a bit like saying that a self-driving car doesn’t need roads. It just doesn’t make sense. Yet with the European Commission’s Omnibus proposals sparking debate over regulatory simplifications, there are suggestions that companies don’t need to digitise their reports — perhaps AI can extract financial and sustainability data directly from unstructured reports instead. It’s wishful thinking: if we want markets that score high on trust, comparability and investor interest, we need clear, accessible and transparent digital disclosures.
What AI can do is enhance analysis, help facilitate tagging as a “co-pilot” and help improve reporting consistency. The latest blog from the We Mean Business Coalition offers some great insights on the role of AI in digital reporting. Authors Jane Thostrup Jagd and Sebastian Fischbach have published an in-depth report on data quality in sustainability data – also on this week’s reading list! – so we were interested to get their quality-focused perspective.
As they reflect, AI is a game-changer for helping companies with straightforward XBRL tagging, detecting errors, and enhancing analytics, but it is not a magic wand. AI models work probabilistically, meaning they learn patterns and make predictions but cannot comprehensively capture exactly what a company seeks to get across to users. XBRL ensures that disclosures follow standardised formats, making financial and sustainability data traceable, comparable, and auditable – all essential for investors, regulators, and capital markets.
AI can play a valuable role in facilitating and improving the reporting process. AI-powered XBRL tagging tools are already improving speed and accuracy, reducing the reporting burden for companies. But they’re not perfect. Human oversight is needed to prevent errors. The real power lies in the combination: AI accelerates the process, handling perhaps 80% of the work, but management oversight ensures the final 20% captures nuances AI can’t.
The result? A single, authoritative version of the truth – ready for both human and AI-driven analysis. That’s why XBRL matters.
Dive deeper here.
