AI-generated legal claims: is your business prepared?

Generative AI has changed what it means to be a self-represented litigant. AI tools allow individuals to quickly generate large volumes of complaints, pleadings, correspondence and court filings.

Legal systems are now facing an increasing volume of AI-assisted filings - some well-founded, some not. This is particularly an issue in jurisdictions that intentionally facilitate individuals or unrepresented parties to bring low-cost proceedings, including employment dispute resolution forums (such as the Fair Work Commission) and administrative appeal tribunals.

Courts, tribunals and businesses will need to adapt to respond to this increase in disputation and proceedings generated by AI-assisted parties.

The Nippon case: a cautionary illustration

A recent US lawsuit, Nippon Life Insurance Company of America v OpenAI Foundation, highlights some of the challenges that arise when AI tools encourage self-represented individuals to pursue legally inappropriate action.

Nippon had previously settled a long-term disability claim with a former employee, and the proceedings were dismissed. Nippon’s complaint alleges that the former employee later pasted her lawyer’s settlement email and court documents into ChatGPT. ChatGPT allegedly characterised the settlement as unfair, encouraged the user to terminate her legal representation, and generated tailored motions seeking to reopen proceedings that had already been dismissed, along with follow-up filings that were ultimately rejected as baseless.

Nippon is suing OpenAI for approximately USD 300,000 in defence costs and USD 10 million in punitive damages, based on three headline causes of action under Illinois law:

• tortious interference with contract, on the basis that ChatGPT induced or facilitated breach of a settlement agreement by encouraging the pursuit of claims already agreed to be finally released;
• abuse of process, on the basis that ChatGPT induced repeated and allegedly meritless filings that imposed cost and burden on the counterparty; and
• unauthorised practice of law, on the basis that ChatGPT crossed the line from providing general information to providing tailored, personalised legal assistance and drafting.

The case illustrates, in concrete terms, the burden that AI-assisted filings can impose on counterparties: multiple rounds of baseless filings, significant defence costs, and a protracted dispute that a straightforward settlement agreement was designed to prevent. While AI provider liability remains untested in Australia, the pattern of conduct described is not unique to the United States.

Strain on Courts and businesses

When unrepresented parties use AI to engage in legal proceedings, legal experts are present to contextualise or appropriately qualify the output claims.

The result is a system under strain. Adjudicative bodies are absorbing an increased volume of filings, some of which lack merit, while, in addition to engaging in resolution of legitimate disputes, businesses are required to dedicate resources to responding to claims that lack merit or might otherwise have been more effectively resolved through standard complaints handling processes.

Resourcing pressure on decision makers and counterparties

Unrepresented parties are able to use AI to generate large volumes of materials very quicky and potentially with no or only limited oversight regarding the legal or factual accuracy of such materials.

Businesses and decision-makers must dedicate time and resources to engaging with such materials, regardless of whether or not these materials are accurate or contain merit. This is particularly acute for consumer-facing businesses, where consumers may now be more likely to institute proceedings rather than seeking to resolve complaints through a business’s complaint resolution processes.

The President of the Fair Work Commission has reported significant increases in workload associated with AI-assisted filing behaviour. This has introduced a new problem - one that impacts access to justice in a different way, by overwhelming adjudicative bodies with materials and/or claims that lack have merit.

Risk to consumers

Consumers relying on AI-generated outputs may not be advised on all available options or the risks associated with each course of action. Consumers seeking recourse regarding a dispute risk:

• exposure to cost orders in relation to proceedings; or
• failure to obtain a resolution that could more efficiently and quickly have been made available through complaint handling processes.

Institutional Responses: disclosure and case management

Courts and tribunals are beginning to respond through disclosure expectations and case management rules.

For example, the Federal Court of Australia has recently issued a practice note stating that the Court may require parties to disclose whether and how generative AI has been used in a proceeding. The Fair Work Commission has also recently published an exposure draft guidance note requiring parties to disclose if AI was used to prepare lodged documents, with explicit consequences for non-compliance (including potential adverse costs orders).

These developments reflect a growing institutional awareness of the challenge, but they stop short of addressing how businesses - particularly those frequently appearing in consumer-facing dispute resolution forums - should structure their own response.

Practical steps for businesses

There are practical steps that can be taken now to manage the impact of AI-generated materials from unrepresented complainants.

A central focus should be the complaints-handling process. Where issues can be identified and resolved early - before they are escalated into formal legal proceedings - businesses may be able to reduce the burden of responding to AI-generated materials from unrepresented complainants.

In practice, this may involve the following types of measures:

• Earlier triage of complaints: ensuring frontline teams can quickly identify complaints that are likely to be escalated, including those supported by AI-generated materials. These processes should facilitate rapid internal escalation and early resolution of higher-risk complaints before they enter formal dispute channels.
• Focusing on key issues: businesses can review complaints with the potential use of AI front of mind and aim to focus dispute resolution efforts on what appear to be the key issues being raised by the complainant. In some cases, it may be helpful to expressly engage with the complainant in this way – e.g. refer to their 50-page complaint, note that the key concerns appear to be X and Y and explain the organisation's position on those matters.
• Addressing root causes: businesses should utilise customer feedback to rapidly identify root causes in their products or processes that are generating complaints and seek to address these pain points for all customers, rather than getting stuck in a stand-off with any particularly agitated complainants.
• Efficient response workflows: developing processes and materials (including templates, workflows etc) to facilitated efficient, targeted and consistent responses to AI-assisted claims, particularly in high-volume forums such as administrative tribunals and the Fair Work Commission.