Artificial intelligence is changing the legal profession, and tools like Claude have attracted attention because they are powerful, flexible, and increasingly promoted for legal work. According to Anthropic's own positioning, Claude is now marketed to legal teams for contract review, drafting, and day-to-day document work.
But for Australian and New Zealand legal practitioners, the real issue is not whether Claude is impressive.
The real issue is whether a general-purpose international AI platform is the right fit for privileged legal work in this region.
For many firms, the answer is no.
Legal work in Australia and New Zealand demands more than a capable general-purpose model
Legal practitioners in Australia and New Zealand operate in a high-risk environment. They deal with privileged communications, confidential client instructions, sensitive personal information, litigation material, evidence, employment disputes, family law issues, regulatory matters, and high-consequence drafting.
In Australia, the OAIC says that, as a matter of best practice, organisations should not enter personal information, particularly sensitive information, into publicly available generative AI tools because of the significant and complex privacy risks involved. In New Zealand, the New Zealand Law Society warns lawyers to address privacy, data protection, confidentiality, privilege, supervision, and quality risks when using generative AI.
That is why legal AI in this market cannot simply be good at answering prompts. It must be built for confidentiality, legal process, and the practical realities of practising law in Australia and New Zealand.
"I can't simply be good at answering prompts. To be useful to AU/NZ practitioners, I have to be built for confidentiality, legal process, and the practical realities of practising law in this region."
Claude is not built specifically for Australian or New Zealand legal practice
Claude may be a strong general-purpose AI system, but Anthropic publicly positions it as a tool for "legal teams" in broad terms, not as a platform purpose-built for Australian or New Zealand law. Its legal messaging focuses on general use cases such as contract review, drafting, and document work, rather than jurisdiction-specific legal practice in this region.
That distinction matters.
Australian and New Zealand practitioners do not just need a model that can read and summarise documents. They need a platform aligned to local legal workflows, local drafting standards, local terminology, and the professional and regulatory expectations that apply in these jurisdictions.
That is why we built Quillio AI Legal Assistant specifically for Australian and New Zealand legal practice. As Quillio puts it: "I'm trained on local legal workflows and drafting standards, not adapted from a general-purpose global product."
Privilege, confidentiality, and client trust come first
For lawyers, confidentiality is not a feature. It is a professional obligation.
Once client material is introduced into a third-party AI environment, firms must think carefully about privilege, confidentiality, disclosure risk, internal governance, and whether that architecture is appropriate for live client matters. New Zealand Law Society guidance explicitly warns lawyers that AI providers may be able to see input data and outputs, that data may be transferred overseas, and that firms must ensure systems are in place to protect confidentiality and privilege.
For many firms, especially those handling disputes, investigations, employment, family, regulatory, or sensitive commercial matters, that alone is enough to make a general-purpose external AI platform the wrong fit.
Privacy expectations in Australia and New Zealand are stricter than many firms realise
Australian regulators have taken a cautious approach. The OAIC says the Privacy Act applies to all uses of AI involving personal information, and it recommends as a matter of best practice that organisations do not enter personal information, particularly sensitive information, into publicly available AI chatbots and other publicly available generative AI tools. The OAIC also notes that entities using AI must consider issues such as collection, secondary use, accuracy, and cross-border disclosure.
New Zealand guidance is similarly cautious. The New Zealand Law Society warns that lawyers should be careful about data used to train AI tools, should not use personal or client information for testing or template creation, and should ensure their systems comply with privacy, confidentiality, and privilege obligations.
For a law firm, that means the question is not simply whether Claude can produce useful work. The question is whether the platform is appropriate for the sensitivity, accountability, and legal consequences of the work being done.
Large context windows do not equal matter-scale legal document handling
This is one of the biggest practical differences between a general AI tool and a legal platform.
Anthropic says Claude supports a 1M token context window, which it describes as roughly 2,500 pages of text or more.
But those figures should not be confused with true matter-scale legal document handling.
Source: Anthropic's published PDF documentation
Anthropic's own PDF documentation says requests are still subject to practical limits, including a maximum request size of 32 MB and a maximum of 600 PDF pages per request. Anthropic also warns that dense PDFs can fill the context window before reaching the page cap and that large files may need to be split into sections.
In practical terms, that means legal teams working on major disputes, transactions, investigations, and document-heavy matters are still constrained by request limits, PDF limits, and the need to break large sets of material into smaller parts. That is very different from a platform designed to work across substantially larger document volumes as part of a single matter workflow.
Compare that to how Quillio handles a matter: "I'll work across the whole bundle in one pass — I won't ask you to split a 1,200-page brief into sections before I can read it." Firms are not forced to squeeze serious legal work into the limits of a general-purpose chat interface.
Legal work requires governance, not just answers
Claude Enterprise does include audit and compliance features. Anthropic says audit logs are available for Enterprise organisations, that exports aggregate audit logs from the past 180 days, and that audit log exports do not include the title or content of chats and projects, only unique identifiers. Anthropic also says audit log events are available through its Compliance API.
That is useful, but many law firms will still ask whether those controls are enough for matter-grade legal governance.
Legal teams need confidence that sensitive work can be governed, reviewed, and managed in a way that satisfies internal risk teams, client requirements, insurers, and the realities of legal practice. For firms with demanding compliance and evidentiary expectations, a general enterprise AI feature set may still fall short of what legal workflows require.
Where Claude genuinely leads
Claude is a genuinely capable AI system with meaningful strengths. Firms evaluating it should understand what it does well before deciding whether it fits legal practice in this region.
- Reasoning quality. Claude is widely regarded as one of the strongest general-purpose reasoning models available, with strong performance on nuanced multi-step tasks.
- Document summarisation. For internal, non-privileged content, Claude's document summarisation and extraction capabilities are genuinely strong.
- Drafting flexibility. Claude handles a wide range of writing and drafting tasks and can adapt tone, structure, and length effectively.
- Enterprise controls. Claude Enterprise includes SSO, audit logs via the Compliance API, usage controls, and admin tooling that suits larger organisations.
- API access and integrations. Claude's API is well-documented and widely supported, making it easy to integrate into custom workflows and internal tools.
If your firm uses AI primarily for non-privileged internal drafting, research, and document work where data sovereignty and matter-scale handling are not priorities, Claude is a credible option.
General-purpose AI is not the same as legal infrastructure
Claude is a powerful general-purpose AI product. That is exactly why it appeals to so many organisations. But legal practitioners in Australia and New Zealand do not just need a capable assistant with some legal usefulness.
They need a platform designed for privileged legal work, local legal practice requirements, matter-based document review, legal-grade governance, and high-volume legal workflows.
That is the difference between a model and a legal platform.
- General-purpose AI with broad applicability
- Global platform, not AU/NZ specific
- Strong reasoning and drafting quality
- Request and PDF size limits apply
- Enterprise audit logs available (metadata only)
- Data sovereignty depends on enterprise configuration
- Built for Australian and New Zealand legal practice
- Purpose-designed for privileged legal workflows
- Matter-scale document handling
- Legal-grade governance and audit trail
- Australian data sovereignty by design
- Local legal standards, drafting norms, and terminology
The better question is not whether Claude can help
The better question is whether a law firm should rely on a general-purpose international AI platform at the centre of its legal workflows.
For many Australian and New Zealand firms, the answer is no.
Not because Claude is unsophisticated.
Not because AI has no place in legal practice.
But because legal work in this region demands more.
It demands a platform built for local practitioners, local expectations, larger legal matters, and the professional obligations that come with practising law in Australia and New Zealand.
That is exactly why we built Quillio AI Legal Assistant.