“AI is increasingly a system-risk story hidden inside a growth story”, says partner Nick Valentine
Sixty-nine percent of senior decision-makers in Australia and New Zealand have raised that parties in their AI value chains are seeking to shift liability for AI errors, according to The AI Year report published by DLA Piper.
As a result, recourse and accountability in such instances are unclear. Moreover, 60% of ANZ-based respondents reported that their commercial strategies hinged on AI suppliers they do not manage or have audit authority over.
Eighty percent of respondents indicated that AI suppliers often could not fulfil explainability and transparency requirements; 71% expressed concern regarding how only a few dominant operators control vital components in the AI value chain. Sixty-one percent said that regulatory controls on compute and data location, including export controls and localisation rules, drove them to halt or rethink AI projects in the past 12 months.
“AI is increasingly a system-risk story hidden inside a growth story. The headline is confidence. But beneath that confidence, there are clear signs of operational strain,” Nick Valentine, New Zealand head of data and technology at DLA Piper, told NZ Lawyer. “Australian and New Zealand businesses are identifying material risks like supplier concentration, limited audit rights, transparency gaps, liability shifting, cyber risk and infrastructure dependency”.
Nonetheless, he noted that just 24% of ANZ respondents that had adopted AI considered auditing supplier and partner governance standards as a key strategic priority this year.
“This is not a story of leaders missing the risks. It is a story of leaders seeing the risks, but that awareness is not always translating into an equivalent level of governance action. That suggests the risks businesses are facing are not unknown but are not always being matched by the governance response. Ultimately, that creates, in my view, a confidence gradient, where strategic optimism outpaces some aspects of operational readiness”, Valentine told NZ Lawyer.
He explained that legal teams need to help their businesses understand “where control actually sits in the AI value chain”.
“That means advising on acceptable use cases, assessing supplier and concentration risk, understanding data and IP implications, documenting decisions, and ensuring there are clear accountability and escalation pathways when issues arise”, Valentine told NZ Lawyer. “In practice, a pragmatic legal approach is not to assume that every supplier risk can be negotiated away. Where contractual leverage is constrained, legal teams can help the business make informed risk decisions, build governance around the constraints of the market, and ensure that AI adoption remains accountable and resilient”.
He added that legal teams needed to “be realistic about where leverage sits in the AI ecosystem”.
“For many New Zealand organisations, particularly when dealing with large global AI providers, contractual leverage may be limited. They may not have the scale or purchasing power to negotiate meaningful changes to standard terms, obtain bespoke audit rights, or secure the level of transparency and explainability they would ideally like”, Valentine told NZ Lawyer. “That does not mean governance stops at the contract. In fact, it makes governance more important. Where organisations cannot negotiate control, they need to build compensating controls around supplier selection, risk assessment, testing, monitoring, data governance, security, and human oversight”.
For The AI Year report, DLA Piper surveyed 975 senior decision-makers like CXOs, senior leaders and directors across 13 markets and 5 sectors.
