Why AI matters for Universal Acceptance and IDNs

2 July 2026, Alfredo Calderón-Serrano.

AI can accelerate Universal Acceptance (UA) and IDN adoption by making multilingual domain names easier for end users to discover, test, translate, explain, and troubleshoot. But it can also increase trust and security risks, especially through AI-generated phishing, look-alike domains, automated domain abuse, and uneven AI performance in low-resource languages.

The central conclusion is this: AI will not solve UA by itself. AI can improve the user experience around multilingual domain names, but only if the underlying systems—forms, email platforms, browsers, identity systems, CRMs, government portals, and procurement rules—are actually UA-ready.

Current baseline: why AI matters for UA and IDNs

Universal Acceptance means that all valid domain names and email addresses, including new TLDs, IDNs, and internationalized email addresses, should work across Internet-enabled applications, devices, and systems. ICANN frames UA as a gateway to the “next billion Internet users” and a more linguistically diverse online experience. (ICANN)

The IDN infrastructure has advanced. ICANN’s 2025 IDN report states that, as of June 2025, 151 IDN TLDs had been delegated, representing 37 languages across 23 scripts; ICANN also reports more than 11,000 IDN tables in the IANA Repository. Under gTLDs, Chinese-script IDN registrations account for 49%, followed by Latin-script IDNs at 28%. (ICANN)

The adoption problem remains at the user-experience layer. UASG’s 2025 readiness materials report that UA awareness and adoption efforts include 18 UA Ambassadors, six local initiatives, and more than 50 UA Day events each year, reaching thousands of people. However, a 2025 ICANN83 community update also noted that only around 14% of websites were UA-enabled and that roughly 26–27% of email systems were fully or partially UA-ready, indicating substantial remaining implementation gaps. (uasg.tech)

The stakes are large. ITU estimates that in 2025 about 6 billion people are online, while 2.2 billion remain offline; the report stresses infrastructure, affordability, and digital skills so that people can benefit from technologies such as AI. (ITU) For many of those current and future users, language and script are not cosmetic issues. They are part of identity, access, trust, and participation.

Figure 1: Created: Why AI matters for UA and IDNs. ChatGPT, 22 June 2026 (c) ACS

Positive impacts of AI for end users

Area	How AI can help	End-user benefit	Evidence / data
Awareness	AI assistants can explain IDNs, UA, EAI, and safe use in local languages.	Users learn that local-language domains and emails are legitimate options.	UNESCO and ICANN are promoting UA through policy development, public procurement, digital services, capacity-building, and awareness-raising. (UNESCO)
Usability testing	AI can automatically test forms, apps, and workflows using IDNs and internationalized emails.	Fewer “invalid email” or broken-form experiences.	UASG identifies UA as requiring systems to accept, validate, process, store, and display domain names and email addresses correctly; AI can support automated QA around those functions. (IANA Lists)
Multilingual support	AI translation and language interfaces can help users understand domains, email addresses, warnings, and help documentation.	Lower barrier for users who do not primarily use English.	UNESCO frames UA as essential to an inclusive multilingual digital environment where domain names and email addresses work regardless of script or language. (UNESCO)
Developer support	AI coding assistants can detect outdated regex validation, ASCII-only assumptions, and poor Unicode handling.	More websites and apps become UA-ready faster.	ICANN calls on developers and system administrators to ensure applications support UA and IDNs. (ICANN)
Security education	AI can produce localized safety guidance about IDN spoofing, phishing, and trusted domains.	Users can learn safer habits in their own language.	IDN Implementation Guidelines 4.1 went into effect in 2025 to strengthen protections against consumer confusion and DNS abuse. (ICANN)

Practical example

An AI-enabled browser or government portal could help a user understand why usuario@ejemplo.empresa, 用户@例子.公司, or مريم@مثال.إختبار is valid, while also warning when a visually similar look-alike domain may be suspicious. That combines inclusion with security, which is exactly where UA and AI should meet.

Negative impacts and risks

Risk	Why it matters for end users	Supporting evidence
AI-generated phishing	AI can make phishing messages more fluent, localized, and persuasive, especially in the user’s own language.	The UK National Cyber Security Centre assessed that AI will increase the effectiveness, volume, and impact of cyber operations over the near term. (National Cyber Security Centre)
IDN homograph abuse	Attackers can combine multilingual characters with AI-generated messages to impersonate trusted brands or institutions.	Research on IDN homograph attacks shows that visually similar Unicode characters can be abused to create look-alike URLs; ShamFinder studied IDN homographs in the wild. (arXiv)
Automated domain abuse	AI can help attackers generate convincing domain names at scale, including phishing and squatting domains.	PhishReplicant detected 3,498 attacker-acquired generated squatting domains in a four-week experiment, with 2,821 used for phishing within a month. (arXiv)
Low-resource language inequality	AI systems often perform worse in low-resource languages, which could reproduce the very exclusion UA is meant to address.	Stanford HAI reports that most major LLMs underperform for non-English and especially low-resource languages, are less culturally attuned, and are less accessible in parts of the Global South. (Stanford HAI)
Moderation and safety gaps	AI-based safety tools may fail in languages with limited training data.	A 2025 study of moderation pipelines for Tamil, Swahili, Maghrebi Arabic, and Quechua found that systems designed around English often fail to account for linguistic complexity in low-resource languages. (arXiv)
False sense of readiness	AI may translate or explain a domain correctly while the underlying system still rejects the IDN or EAI address.	UASG’s roadmap emphasizes actual deployed technology—websites, applications, email tools, and services—as the measure of UA readiness, not just awareness or documentation. (IANA Lists)

Impact assessment: likely outcomes for end users

Positive scenario

AI becomes a UA accelerator. It helps developers test systems, helps public agencies procure UA-ready platforms, helps users understand local-language domains, and strengthens phishing detection in multiple scripts. In this scenario, AI supports ICANN and UNESCO’s policy direction: procurement, public services, capacity-building, and awareness. (UNESCO)

Negative scenario

AI becomes a trust destabilizer. Attackers use generative AI to create more convincing multilingual phishing campaigns, register deceptive domains, and exploit users who are newly encountering IDNs. Meanwhile, legitimate services continue rejecting valid internationalized emails and domains, reinforcing user distrust.

Most realistic scenario

Both happen at once. AI will improve UA testing, awareness, and multilingual support, but it will also increase abuse pressure. The decisive factor will be governance: whether governments, universities, registries, registrars, platforms, and vendors treat UA-readiness and multilingual security as procurement and compliance requirements.

Figure 2: AI in the multilingual Internet: Opportunity and risk – the future depends on readiness. Created: ChatGPT, 22 June 2026 © Alfredo Calderon

Strategic recommendations

For ICANN, UASG, Internet Society chapters, and civil society, AI should be used to build UA awareness campaigns in local languages, generate user-facing tutorials, and identify websites that fail IDN/EAI acceptance.

For governments and universities, UA-readiness should be included in procurement specifications for identity systems, student portals, CRM systems, email platforms, learning management systems, public service platforms, and cybersecurity tools.

For developers, AI coding tools should be trained or prompted to flag risky patterns such as ASCII-only validation, outdated email regex, hard-coded TLD lists, poor Unicode normalization, and failure to display right-to-left scripts properly.

For end users, education must combine two messages: “Your language belongs on the Internet” and “Multilingual domains still require careful trust checks.”

Bottom line

AI will make UA and IDNs more visible to end users. That is good. It can help people use the Internet in their own language, script, and cultural context. But AI will also make deception more scalable, more localized, and more persuasive.

The future of UA in the AI era depends on whether we connect three layers:

1. Technical readiness: systems accept, validate, process, store, and display IDNs and EAI correctly.

2. Human readiness: users understand and trust multilingual domain names.

3. Security readiness: platforms detect abuse without discouraging legitimate multilingual use.

The goal is not simply an Internet that can display every script. The goal is an Internet where every user can safely and confidently use their language as part of their digital identity.

---

References

• ICANN. (2025). ICANN highlights IDN progress with release of IDN annual report June 2025. Internet Corporation for Assigned Names and Numbers.

• ICANN. (n.d.). Universal Acceptance (UA). Internet Corporation for Assigned Names and Numbers.

• ITU. (2025). Facts and Figures 2025. International Telecommunication Union.

• National Cyber Security Centre. (2024). The near-term impact of AI on the cyber threat. UK NCSC.

• Stanford HAI. (2025). Mind the language gap: Mapping the challenges of LLM development in low-resource language contexts.

• Universal Acceptance Steering Group. (2025). UASG 10 Years and 2025 UA Readiness Report.

• UNESCO. (2026). Universal Acceptance. United Nations Educational, Scientific and Cultural Organization.

About the Author

Alfredo Calderón-Serrano is a Puerto Rican educator, instructional designer, and Internet governance leader with more than three decades of experience in higher education, educational technology, and capacity building. An active At-Large/NARALO volunteer with ICANN since ICANN53, he served as the 2024 NomCom delegate for the North American At-Large community and is co-founder of the North American School of Internet Governance (NASIG) and the Virtual School on Internet Governance (VSIG). In 2025, the ICANN Board awarded him the Dr. Tarek Kamel Award for Capacity Building in recognition of his work founding and sustaining these schools. A board member of the Internet Society Puerto Rico Chapter since 2014, he focuses on Universal Acceptance, IDNs, accessibility, and digital inclusion — translating complex governance topics into learning opportunities for academic, civil society, and regional communities.