Plain language and multilingual intake
Agent intake that works across literacy levels and languages: plain language by default, an honest indicator of how well the agent handles the user's language, and a human interpreter when it doesn't. That disclosure stops the service from settling consequential matters in a language the agent only appears to understand.
The impact of agents
As more government interaction runs through AI agents, the same capability that could open bureaucratic language to users with lower literacy or limited English proficiency can instead deepen the barrier. It rewards users who know how to prompt the agent well and underserves languages with little training data behind them. Left unaddressed, the agent's own language competence decides who is understood, without anyone having chosen that outcome.
What must be verified
Government needs confidence that a user acting through an agent has understood the service and been understood by it, regardless of their literacy or the language they use. Meeting that requires plain language as the working register, disclosure of how well the agent commands a given language, and human-verified translation where legal or medical content makes an error costly. The agency or agent provider must make that disclosure and commission the verified translation.
Protecting access
An agent fluent in English and unusable in Karen or Dinka excludes speakers of under-resourced languages without refusing them anything or recording a failure: it delivers confident, wrong sentences the user has no way to check.
Keeping the path open
- Make plain language the default register, rather than an option the user must find.
- Confirm the detected language with the user at intake.
- Show a competence indicator when the agent's command of a language is limited.
- Keep a human interpreter one action away, including for the legal or medical content that already requires human-verified translation.
Response surface
The agent states how well it handles the user's language and offers a human interpreter where its command of that language runs out.
Which language should we use?
Your browser is set to Vietnamese, so it’s selected. Change it if that’s wrong. Plain language is the default in every language.
The support level is set by accredited translators and republished as it changes, so it is an audited rating rather than the system’s own estimate of how well it is doing.
Maturity
- Established
For plain-language requirements, codified and in force.
- Emerging Headline
For multilingual AI with quality transparency, which is beginning to appear.
- Frontier
For language-quality disclosure patterns on agents, which have no established precedent.
Precedents
The US Plain Writing Act. The Act requires executive branch agencies to use plain language in documents the public needs in order to obtain benefits, access services, or comply with requirements. Agencies must train employees, establish compliance oversight, write all new or substantially revised documents in plain language, and publish an implementation plan. Plain language is a statutory obligation on the agency, and not a style preference.
Executive Order 13166 on limited English proficiency. The order required federal agencies to provide meaningful access for people with limited English proficiency, through language-access plans, staff training, multilingual recruitment, qualified translators and interpreters, and language-assistance technology. Executive Order 14224 declared English the official language and revoked it. The underlying legal requirements for language access remain in force.
TIS National (Australia). The Translating and Interpreting Service provides language services for people with limited English proficiency and for the organizations that support them, supporting the Multicultural Access and Equity Policy. Australia's framework, developed from the Lo Bianco report, is built on English-plus multilingualism and on removing language-based social inequalities. A standing national interpreting service is the infrastructure the policy rests on.
What carries over to agent use
High, with one caveat. AI agents can do better than static document translation because they adapt to the user's language level. That creates a new dependency, though: the quality of multilingual AI output varies widely by language. Stanford research (2025) documents how LLMs leave non-English speakers behind. The "invisible languages" problem (languages with insufficient training data) means agent-mediated services could be excellent in English and Mandarin but unusable in Karen, Dinka, or Auslan.
The plain-language and interpreter obligations transfer directly. The disclosure does not: none of the cited regimes require a service to say how well it commands the language it is operating in, so the competence indicator has no precedent behind it.
Where things go wrong
Where this goes wrong is a user misled by confident-but-wrong agent output into a position that harms their entitlement or compliance. A service can also claim multilingual support in aggregate while specific languages (Karen or Dinka rather than Spanish or Mandarin) stay effectively unsupported behind the combined number. Disclosing competence per language rather than as one average, and defaulting to plain language, addresses both.
Sources
8 references
The instrument, the operating deployment, or the official record itself.
Writing about the subject rather than the framework itself, including vendor commentary.
- Digital.gov, "Requirements for improving access to services for LEP" (describes EO 13166, revoked March 2025)
- Australian Academy of the Humanities, "Enshrining multilingualism"
- US Department of Labor — EO 13166 24th anniversary release
- Migration Policy Institute — 'Shifting Priorities' on the official-English order