💻 Key AI CONTRACT Clauses You Can’t Ignore

The AI boom has sparked countless discussions on its benefits, risks, governance, and global regulations. But as AI adoption accelerates, a critical question arises for New Zealand organisations: Are existing contract templates fit for purpose when procuring AI-powered IT services?

💥 The Myth of the ‘Standard AI Clause’ 

Many organisations are seeking a ‘one-size-fits-all’ AI clause. The reality is that no single provision can cover the diverse risks and uses of AI. Instead, contract drafters must take a nuanced approach, balancing legal safeguards with practical considerations.

💢 Common Pitfalls to Avoid

💡 Over-reliance on Generic Requirements – Standard ICT contracts often require compliance with laws and good industry practice. However, New Zealand lacks AI-specific regulation, making such clauses inadequate. With AI evolving rapidly, defining ‘good industry practice’ is also a challenge.

💡 Blind Use of Model Clauses – AI-specific contractual templates exist, such as the European Commission’s AI clauses or the UK’s Society for Computers and Law samples. While useful, overly prescriptive clauses can complicate negotiations. For instance, demanding full transparency on an AI tool’s workings may be unrealistic if proprietary algorithms are involved.

💢 To draft effective AI clauses, start with these core issues:

💡 Consent to AI Use – Should a supplier need approval to integrate AI into an existing service? If impractical, customers might instead require notification and the right to exit contracts if AI integration is unacceptable.

💡 Transparency – What level of disclosure should suppliers provide on how AI tools operate? Customers may require clarity on AI decision-making processes.

💡 Monitoring – Should human oversight be mandated? If so, how can suppliers demonstrate effective monitoring?

💡 Data Usage & Ownership – Does AI training involve customer data? Can that data be removed? Contracts should ensure definitions of ‘data’ and ‘confidential information’ cover AI-generated content.

💡 Intellectual Property (IP) – Who owns AI-generated outputs? What protections exist against IP infringement claims from AI-generated content?

💡 ‘Kill Switch’ Provision – Can customers disable AI functionality if needed? If so, what impact does that have on service delivery?

💡 Adaptability to Change – Given evolving AI regulations, should contracts allow termination or modification rights to respond to new laws?

💡 Liability & Risk Allocation – What happens if AI fails? Who bears responsibility? Liability must be carefully assessed against practical risk scenarios and insurance coverage.

AI’s integration into IT contracts demands tailored solutions, not generic templates. By addressing these key concerns upfront, organisations can navigate AI procurement with confidence.