Firms choosing AI tools must ask whether every lawyer and client can actually use them
As more law firms adopt AI, accessibility belongs within the professional obligations firms weigh: competence, supervision, client service and accommodation. A tool’s accuracy and confidentiality protections are essential, but firms must also establish whether the lawyers and clients expected to rely on it can actually use it.
When the tool works but the interface doesn't
Imagine a law firm introducing an AI research assistant. The demonstration is impressive: questions produce answers in seconds, with links to supporting authorities. A blind associate can read the answer with a screen reader, but cannot navigate the panel containing the sources. The firm expects every lawyer to verify the output. Its chosen tool makes that harder for one of them.
In my article on using AI-powered recruitment platforms, I examined how these platforms may be liable for discrimination. The question now extends into everyday practice: what happens when the technology a firm selects becomes a barrier to doing legal work or obtaining legal help?
As a lawyer with profound hearing loss, I know that offering another way to communicate helps only if that alternative is itself accessible. A telephone number does little for someone who cannot use the phone. The same principle should guide firms as they introduce AI into research, drafting, intake, and client communication.
Consider an intake system that requires spoken answers but offers no workable text option for a client with a speech disability, a drafting platform whose review controls cannot be operated without a mouse, or a client-intake chatbot that times out before someone using assistive technology can finish, losing the very information the firm needs.
A system can generate useful content while remaining inaccessible to the person who needs it. Firms must examine both the interface and the documents it produces.
That distinction matters when an AI-generated summary moves from an internal workspace into a client’s hands. A document that looks orderly on screen may lack the structure a screen reader needs to navigate it. Firms should check the delivered version, including headings and reading order, and ask clients which formats work for them. Rewriting advice in plainer language can help, but someone must still confirm that the legal meaning survives.
Accessibility as a professional and career issue
For lawyers, the implications reach beyond inconvenience. If access to a firm’s preferred tools affects who receives work assignments, how quickly work is completed or whose performance appears strongest, an accessibility barrier can become a career barrier. Before attributing a productivity gap to an associate, a firm should ask whether its systems gave them an equal opportunity to work.
There is also a connection to professional judgment. Technological competence includes understanding the technology relevant to one’s practice; the Federation of Law Societies of Canada’s Model Code addresses this in its commentary to rule 3.1-2, though the applicable rules remain those of each jurisdiction. Can the lawyer inspect the sources, review the changes, and identify errors using the available tools? Human oversight depends on the ability to access what requires oversight.
For clients, the consequences may arise before a lawyer sees a file. An inaccessible intake process can prevent someone from explaining their problem; information sent in an unusable format can prevent a client from understanding advice or giving informed instructions.
Existing human rights obligations already matter. In Ontario, employers and service providers must accommodate disability-related needs to the point of undue hardship, and the Ontario Human Rights Commission’s policy emphasizes inclusive design and the prevention of new barriers. A vendor’s assurance does not determine whether a particular lawyer or client can use the service; firms should weigh accessibility when selecting technology, while remaining ready to accommodate individual needs.
What evaluation looks like before a contract is signed
The practical work begins before a contract is signed, while a firm can still compare products and negotiate. A polished demonstration cannot establish whether a product works with the assistive technology its users rely on. Meaningful evaluation follows a complete task – signing in, inspecting citations, revising, and exporting a document in an accessible format. Testing with people with disabilities, with appropriate support and compensation, reveals barriers a demonstration misses.
Training deserves the same scrutiny as the product. A lawyer may be able to use a platform yet be excluded from learning how it works if demonstrations are uncaptioned or instructions depend entirely on visual cues. Firms should make training materials accessible and allow staff to practise with their own assistive technology before the tool becomes part of expected daily work.
Findings should shape both the contract and the deployment decision. Accessibility expectations, responsibility for defects, and a process for update-related barriers should be explicit, and someone must have the authority to pause deployment when an essential task is inaccessible. Alternatives should preserve timely service, confidentiality, and independence. Requiring a client to rely on a family member, an associate, or a colleague can undermine those safeguards.
Maintaining accountability after launch
Responsibility must continue after launch. Firms should identify who receives reports of barriers, who arranges an immediate alternative, and who follows up with the vendor. A client facing a deadline cannot wait for the next software release. Recording recurring problems also helps a firm decide whether a product remains suitable when its contract comes up for renewal.
AI can benefit people with disabilities, making careful selection even more important. Firms have an opportunity to improve how legal work is done and how clients reach them – provided accessibility is part of the decision from the outset.
Before making an AI tool part of everyday practice, a firm should establish that the people expected to rely on it can use it effectively.
