TL;DR
OpenAI has published a page describing how Australian law firm Gilbert + Tobin governs and scales AI with its technology. The available material identifies the initiative but does not disclose its scope, controls, adoption figures or measured results.
OpenAI has published a customer-focused page about how law firm Gilbert + Tobin governs and scales artificial intelligence using OpenAI technology, placing controlled adoption at the center of the initiative. The development matters because legal work involves confidential and privileged information, making governance, professional accountability and data handling central to any wider deployment.
The OpenAI headline identifies two parts of Gilbert + Tobin’s approach: governing AI and scaling its use. It also establishes that the work involves OpenAI. The available material does not identify the products deployed, the legal tasks supported, the number of users involved or the internal rules governing use.
That distinction limits what can be reported as fact. OpenAI is presenting Gilbert + Tobin as a customer example, but the supplied material contains no deployment statistics, performance measures, risk findings or independent evaluation. It also provides no direct statements from the firm, its lawyers, clients or regulators that could establish how the program operates in practice.
The headline’s pairing of governance with scale indicates that oversight is part of the stated deployment story, rather than an issue discussed separately from adoption. It does not establish which safeguards are in place, whether use is mandatory or voluntary, or whether AI-generated work must pass a defined review process before reaching clients.
How Gilbert + Tobin Governs and Scales AI With OpenAI
OpenAI has presented the Australian law firm as a customer example centered on controlled AI adoption. The headline establishes the initiative—but the supplied material does not disclose its products, safeguards, reach, or measured results.
Oversight is part of the stated deployment story, rather than a separate afterthought.
No scope, control framework, adoption rate, quality measure, or financial outcome is supplied.
The announcement should be read as OpenAI’s characterization, not an independent evaluation.
Legal AI operates in a higher-risk environment
Law firms may handle privileged communications, confidential client material, litigation strategy, and advice carrying professional consequences. Wider AI access therefore raises questions that ordinary productivity deployments may not.
Confidentiality
Deployment design must account for what information users may enter, where it is processed, who can access it, and how long it is retained.
Accuracy
Incorrect or unsupported output can create serious consequences when incorporated into advice, court filings, research, or client communications.
Accountability
Lawyers remain responsible for professional work. Clear review requirements and ownership are essential when AI contributes to an output.
What the available material can—and cannot—support
The customer-page framing identifies the organization, the vendor relationship, and the themes of governance and scale. It does not provide enough detail to assess implementation maturity, safety, or business value.
| Question | Status | What is available | Responsible conclusion |
|---|---|---|---|
| Is Gilbert + Tobin using OpenAI technology? | ✓ Stated | OpenAI presents the firm as a customer governing and scaling AI with its technology. | The vendor relationship is part of the stated customer story. |
| Which product, model, or technical configuration is used? | ~ Unknown | No product tier, model, API deployment, or configuration is identified. | No specific OpenAI product should be attributed without further documentation. |
| How is use governed? | ~ Unknown | No approved-use policy, review standard, training program, audit process, or incident procedure is supplied. | The existence or effectiveness of any particular safeguard cannot be verified. |
| How large is the deployment? | ~ Unknown | No user count, practice-area coverage, adoption rate, or deployment timeline is reported. | “Scale” may refer to users, workflows, capacity, or planned expansion. |
| Has the initiative delivered measurable results? | ~ Unknown | No figures for time saved, accuracy, cost, quality, financial return, or client outcomes are available. | Effectiveness and business value remain unestablished. |
What a defensible legal AI program would need to connect
The announcement foregrounds governance, but the operational links below remain undocumented in the supplied material.
Define permitted use
Specify approved tasks, restricted information, eligible users, and acceptable systems.
Control access
Apply permissions, data-handling rules, retention settings, and system boundaries.
Review outputs
Require accountable human verification before AI-assisted work reaches clients or courts.
Measure and improve
Track adoption, errors, incidents, quality, time, and outcomes against defined baselines.
The headline is clear; the operating picture is sparse
This visual is an evidence-availability map, not a performance score. A short bar means the supplied material contains little usable detail about that domain.
The right framing is not yet proof of the right outcome.
Pairing governance with expansion is appropriate for legal AI. But the available headline demonstrates organizational intent, not that Gilbert + Tobin has solved the operational, confidentiality, accuracy, and professional-responsibility risks of generative AI. Confidence should rise only when documented controls and measurable results accompany wider access.
Legal AI Scale Meets Professional Duties
Law firms face a higher-risk adoption setting than many general office users because their systems may process client communications, privileged material and litigation strategy. A program described as both governed and scalable could influence how other firms structure access, review outputs and assign accountability. Yet the headline alone cannot establish effectiveness, cost savings or improved legal outcomes.
The story also shows how AI vendors are using professional-services customers to frame enterprise adoption around managed organizational use. For readers evaluating similar programs, the relevant test is not broad availability. It is whether documented controls and measurable results accompany wider access.
AI governance software for law firms
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Law Firms Face Distinct AI Risks
Generative AI can assist with activities such as drafting, summarizing and searching, but legal organizations must account for confidentiality, accuracy and professional responsibility. Incorrect output can carry consequences when incorporated into advice, filings or client communications. Those risks make human review, access controls and approved-use policies relevant parts of deployment design.
OpenAI’s framing places Gilbert + Tobin within the broader movement toward enterprise-managed AI. The supplied headline does not provide a timeline, prior pilot history or comparison with the firm’s earlier technology systems, so the maturity of the program cannot be established from the available information.
As an affiliate, we earn on qualifying purchases.
Deployment Controls Remain Undisclosed
Several operational questions remain unanswered. The available material does not specify which OpenAI product or model Gilbert + Tobin uses, how many employees have access, which practice areas participate, or whether the system connects to internal knowledge repositories. It also does not disclose data-retention settings, residency arrangements or access permissions.
There is no disclosed information about output review, staff training, audit procedures, incident reporting or restrictions on entering client information. The material also lacks adoption rates, time-saved estimates, error measurements and financial returns. Without those details, readers cannot judge whether “scales” refers to user growth, workflow coverage, technical capacity or a planned expansion.
No direct quotation was available from Gilbert + Tobin or OpenAI representatives. The absence of named speakers and supporting metrics means the announcement should be read as OpenAI’s characterization of the initiative, not as an independent evaluation of its safety or business value.
AI risk management solutions for legal industry
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evidence Needed as Use Expands
The next useful milestone would be publication of deployment scope and governance documentation, including permitted tasks, review requirements and controls for sensitive information. Evidence on accuracy, adoption and client outcomes would also allow a clearer judgment of whether wider use is producing durable benefits.
Further statements from Gilbert + Tobin could clarify who oversees the program, how lawyers are trained and how the firm responds when an AI system produces incorrect or unsupported material. Until then, the scale and results remain undefined.
As an affiliate, we earn on qualifying purchases.
Where I land
My assessment is cautious: pairing governance with expansion is the right framing for legal AI, but a headline cannot show that the controls work. I would treat this as an indication of organizational intent, not evidence that Gilbert + Tobin has solved the operational and professional risks of generative AI.
The strongest counterargument is that law firms may have sound reasons not to publish security controls, client-data arrangements or internal performance details. I accept that some information must remain restricted, but aggregated adoption and quality measures could still support the claims without exposing sensitive systems. I would become more positive if the firm released clear governance principles, review standards and independently checkable outcome data; evidence of recurring errors, weak oversight or client-data exposure would move my assessment in the opposite direction.
Source: OpenAI
Key Questions
What did OpenAI announce about Gilbert + Tobin?
OpenAI published a page describing how Gilbert + Tobin governs and scales AI with OpenAI. The available headline establishes the subject of the customer story, but does not provide operational details.
Which OpenAI products does the law firm use?
The supplied material does not name a model, product tier or technical configuration. It would be inaccurate to identify ChatGPT Enterprise, an API deployment or another product without additional documentation.
How does Gilbert + Tobin govern AI use?
Specific policies are not described in the available material. Details about human review, approved tasks, data handling, audits and employee training remain unavailable, so no particular control can be attributed to the firm.
Has the AI program produced measurable results?
No performance figures were supplied. There are no reported measures for time saved, adoption, accuracy, cost or client outcomes, and no comparison window or baseline from which readers could evaluate improvement.
Why does this development matter to legal clients?
Legal clients may want to know whether AI touches their information or work product and what review applies. Confidentiality, accuracy and lawyer accountability can affect trust, while transparent controls would help clients understand how technology is used in their matters.
Source: OpenAI
