The British Standards Institute (BSI), a body that issues technical standards, has launched a new standard for the developing AI audit market to protect against a “wild west” of unchecked providers.
The BSI warned that while AI assurance is recognised as being crucial to responsible to AI adoption, there are diverging approaches in different countries.
The organisation said that leading accountancy firms, including the Big Four, are launching AI auditing programmes to respond to the growing market demand for transparency.
But to operate effectively certification bodies require standardised methodologies to assess AI systems effectively across jurisdictions.
The Institute said that there is currently a shortage of qualified auditors in the AI market, adding that “hundreds” of firms in the UK offer AI assurance services, many provided by AI developers, which raises concerns about independence.
The organisation describes its new framework, called BS ISO/IEC 42006, as the world’s first international standard designed specifically for certifying bodies that independently audit AI management systems. It aims to provide clear competency frameworks for auditors.
The BSI previously published the first AI management standard in late 2023 and certified businesses to this including KPMG Australia.
The organisation said the new standard will ensure that businesses can be confident that those assessing AI governance are doing so in a clear, consistent and coherent manner.
Additionally, the BSI said its guidance will enable regulators, customers, and investors to differentiate credible AI governance implementations.
Unlike broader AI governance frameworks, the BSI said BS ISO/IEC 42006 is the first international standard dedicated to AI system certification, rather than the AI systems themselves.
This additional layer of verification is designed to strengthen trust in AI governance certifications and ensure consistency among certifying bodies.
Mark Thirwell, global digital director at the BSI said that businesses need to be sure that when their AI management system is being assessed, it is being done in a robust, coherent and consistent manner.
“As companies race to provide AI audit services, there is a risk of a ‘wild west’ of unchecked providers and the potential for radically different levels of assessment,” he added. “The new guidance, BS ISO/IEC 42006 represents a crucial milestone in global AI accountability, by setting clear certification requirements.
“This standard will enable regulators, customers, and investors to differentiate credible AI governance implementations from unchecked claims, supporting responsible AI innovation and paving the way for AI to be a force for good.”
