ChatGPT’s lead in South Korea is more than a headline — it’s the product of distribution, language fit, and local partnerships that have left room for specialist challengers such as Perplexity and SKT’s A.Dot to stake meaningful positions in the market this year. WiseApp·Retail’s November 2025 snapshot puts ChatGPT at 21.62 million monthly active users (MAU) in Korea, with Perplexity at 1.84 million and A.Dot at 1.8 million — a gap that tells two parallel stories: mass-market dominance by a global generalist and rising local traction for research-first and telco-backed alternatives.
Background / Overview
South Korea’s smartphone users have rapidly adopted conversational and generative AI across age groups and use cases, turning these assistants into everyday tools for drafting, research, translation, and quick creative tasks. Industry trackers and local app-analytics firms have been publishing regular snapshots of active users and usage time throughout 2024–2025; those snapshots repeatedly show ChatGPT near the top of usage metrics while regional players and research-focused apps rise in niche slices of activity. A WiseApp·Retail report published in early December 2025 presents the most recent, detailed MAU ranking for November and signals both growth and consolidation in local patterns. This article summarizes the data, verifies the core claims where independent confirmation is available, explains why the ranking looks the way it does, and evaluates the practical strengths and risks for Korean users, enterprises, and Windows-focused IT teams.
What the November 2025 data says
Top-level ranking (WiseApp·Retail, November 2025)
- ChatGPT — 21.62 million MAU.
- Perplexity — 1.84 million MAU.
- A.Dot (에이닷) — 1.80 million MAU.
- Rytune/Wrtn — 1.62 million MAU.
- Grok AI — 980,000 MAU.
- ixi-O — 540,000 MAU.
- Claude — 480,000 MAU.
- Google Gemini — 420,000 MAU.
- Microsoft Copilot — 300,000 MAU.
- DeepSeek — 270,000 MAU.
WiseApp·Retail also identified ChatGPT, Perplexity, Grok AI, ixi-O, and Claude as having recorded their largest user counts ever during November. The survey selected apps with
dedicated interactive generative-AI apps (text, image, personal assistant functions), excluding single-purpose photo, note-taking, or translation tools.
Independent context and corroboration
- The pattern of ChatGPT leading Korean usage is consistent with earlier WiseApp snapshots and Korean business press reports that have tracked rapid and recurring increases in ChatGPT usage and overall generative-AI minutes on device. Historical WiseApp-reported figures (for example, December 2024 and early 2025 snapshots) show ChatGPT repeatedly ranking first in MAU and usage time across age groups. Those prior reports match the structural conclusion: ChatGPT is the dominant consumer-facing gateway to generative AI for many Koreans.
- Broader international confirmation of Korea’s strong ChatGPT market comes from reporting that OpenAI established a local presence in Seoul in 2025 to support partnerships and local demand, underscoring why ChatGPT’s footprint is unusually large in Korea relative to many other markets. That expansion is consistent with the MAU signal reported by WiseApp.
Caveat: the November 2025 MAU breakdown itself is a WiseApp·Retail analysis published by AsiaEconomy. WiseApp’s methodology (sample of Android and iOS users, definition of which apps were included) is explicitly reported by the firm and reproduced in the coverage, but independent third-party app‑telemetry firms may report different absolute numbers depending on methodology (e.g., panel composition, telemetry sources, inclusion rules). Treat the ranking as a robust directional indicator with the caveat that absolute counts will vary by measurement approach.
Why ChatGPT leads in Korea: reach, partnerships, and product fit
Several structural reasons explain ChatGPT’s outsized lead in Korea.
1) Cross‑platform ubiquity and product maturity
OpenAI’s ChatGPT is available across web, mobile apps, desktop clients, and APIs with a mature upgrade path (Free, Plus, higher professional tiers). That multi-surface availability reduces friction for casual users and professionals alike, increasing retention and daily interaction counts. The variety of integrations and third‑party plugins further amplifies utility for content creation, coding, and research tasks — the same everyday functions Koreans report using assistants for.
2) Local partnerships and market investment
OpenAI’s decision to establish a Seoul office and local legal entity reflects a commitment to serve Korean users and partners directly. Local staff and partnerships (OpenAI’s engagements with Korean platform operators) improve language support, regional compliance, and channel distribution, which directly boost user acquisition and subscriber conversions. This regional investment likely contributed to ChatGPT’s unusually high MAU figures in Korea relative to many other countries.
3) Broad use-case coverage (generalist advantage)
ChatGPT functions as a reliable “all‑rounder”: writing drafts, summarizing Korean-language content, ideation, and light coding. That generalist utility tends to win mass-market adoption even when specialist apps outperform it on particular tasks. Users who use a single assistant for most daily tasks will often prefer the tool that solves the most needs adequately.
Why Perplexity ranks second: citation-first search and research workflows
Perplexity’s growth in Korea — and its 1.84M MAU in November — reflects a clear product-market fit that is adjacent to, not directly overlapping, ChatGPT’s strengths.
Perplexity’s unique strengths
- Citation-first answers: Perplexity builds conversational answers with explicit links and source attributions, making it preferable for users who need quick, traceable research rather than open-ended ideation.
- Hybrid model strategy: Perplexity combines its own retrieval and ranking logic with third‑party model families to prioritize evidence and provenance.
- Pro features for heavier users: Professional tiers and integrations that prioritize source visibility and deeper retrieval broaden Perplexity’s appeal for researchers, students, and journalists.
Those design choices explain why Perplexity is often the first stop for verification and citation-aware queries on mobile — and why Korean users with research or source‑driven needs might use it alongside ChatGPT rather than instead of it.
Why A.Dot (에이닷) reaches the top three: telco distribution + multi‑model strategy
A.Dot is SK Telecom’s flagship AI assistant and its inclusion at 1.80M MAU reflects several advantages unique to a telco-backed product.
What A.Dot offers
- Local language conditioning and Korean UX: A.Dot’s models and tuned prompts emphasize Korean language fluency, style, and context, giving it resonance for users who prefer a domestic product experience.
- Telco distribution and bundling: As an SK Telecom product, A.Dot can be bundled into service plans, preinstalled flows, and marketing pushes that accelerate adoption across SKT’s large subscriber base. This distribution channel is a material advantage over global apps that require independent downloads.
- Multi‑model and agent platform: A.Dot positions itself as a multi-model access hub (supporting Perplexity, Claude, and SKT’s own A.X LLM) and focuses on actionable productivity features — meeting summaries, scheduling, and domain agents — which appeal to both consumer and enterprise workflows. SKT’s A.Dot Biz and agent-store concepts move the product toward an “actionable AI” strategy rather than pure chat.
The combination of distribution, localized UX, and enterprise-friendly features explains why a telco product like A.Dot competes effectively with independent research tools and global generalists in the Korean market.
Methodology and data quality — what to accept and what to question
The WiseApp·Retail November snapshot is valuable because it is a consistent, local app-analytics vantage point. However, several methodological realities affect interpretation:
- WiseApp’s sample is a domestic panel of Android and iOS users; different panels, telemetry sources, or inclusion rules (e.g., counting web visits to ChatGPT vs. only app installs) will change absolute MAU counts.
- The survey excluded single-purpose apps (camera, memo), meaning “AI reach” measured here is focused on dedicated generative-AI apps. Some platform-integrated assistants (embedded in kiosks, web browsers or enterprise SSO) may not be fully captured.
- “MAU” is a snapshot of active sessions in a month; it does not measure depth of engagement (minutes per user), paid conversion, or enterprise usage behind SSO/APIs.
- Media replications of WiseApp numbers are reliable for directional reading but should not be treated as the single authoritative truth for market share. Where possible, triangulate with other telemetry providers or vendor disclosures before making procurement or investment decisions.
If your team needs an operational decision (buying enterprise seats, negotiating non‑training clauses, or selecting a default assistant for employees), require vendor-provided telemetry and contract-level assurances rather than relying exclusively on public MAU snapshots.
Strengths and opportunities in the Korean market
- High consumer adoption and diversity of choices: Korean users have rapidly adopted multiple assistants, which fuels experimentation and feature innovation.
- Local players innovate with tight platform integration: A.Dot and domestic offerings excel where integration with local services (payment rails, telco bundles, local content) matters.
- Research-first tools gain utility: Perplexity’s citation-first design answers a real user need — verifiable, quick research — and complements rather than displaces generalist assistants.
- Enterprise-ready variants are appearing: Microsoft Copilot, Claude’s safety‑oriented variants, and enterprise bundles from local vendors offer governance and tenant grounding that larger organizations require.
Risks and blind spots
- Data privacy and residency risks
- Consumer apps may process inputs in foreign cloud regions and have ambiguous training policies. High-profile cases (for example, DeepSeek’s paused downloads due to privacy concerns in Korea) demonstrate how quickly regulatory scrutiny can materialize and pull an app out of distribution. Enterprises and sensitive users should avoid pasting regulated data into consumer tiers unless non-training, data-residency and retention guarantees are contractually provided.
- Hallucinations and provenance
- Generative models remain prone to confident-sounding errors. Perplexity’s citation model mitigates this for research queries, but even citation-first responses require human validation for high-stakes decisions.
- Vendor lock-in and ecosystem lockout
- Embedding an assistant deeply into a platform (workspace, email, telco service) can create operational lock-in. That lock-in brings convenience but increases switching costs and concentrates risk around a single vendor’s policy decisions.
- Measurement and reporting inconsistencies
- Public rankings from single analytics vendors are directional. Differences in panels and counting rules can materially change perceived “rankings.” Procurement decisions should be based on contractual metrics and proofs of performance.
- Monetization and subscription creep
- Free tiers attract users, but the most productive features typically live behind subscriptions. Teams planning organization-wide rollouts must model actual costs (per-seat, per-generation) and guard against unmanaged usage that drives surprise invoices.
Practical guidance for Windows users and IT teams in Korea
Quick rules
- Treat all consumer-tier assistants as drafting tools, not authoritative sources.
- Avoid pasting PHI, PCI, or confidential customer data into consumer apps without a written non‑training guarantee.
- Prefer regional or enterprise plans with tenant-grounding if your org must keep data within a controllable perimeter.
Step-by-step checklist for rolling out AI assistants (for IT managers)
- Inventory current usage: identify which assistants employees use and for what tasks.
- Map data sensitivity: classify tasks by data sensitivity and permit tools accordingly.
- Prefer enterprise-grade connectors: choose assistants that support Microsoft 365, Google Workspace, or local SSO with admin controls.
- Contractually require non‑training clauses or private-instance hosting for regulated data.
- Configure MDM/EDR rules to limit app permissions (camera, full-keyboard access) where unnecessary.
- Pilot with representative teams for 2–4 weeks, track usage and integration pain points, then scale.
- Monitor spend, quotas, and user-reported errors; require human sign-off on any AI-generated legal or financial text.
What to watch next in Korea’s AI market
- Vendor partnerships and bundling moves (for example, telcos offering pro tiers free to subscribers) that can shift adoption quickly. SKT’s bundling of Perplexity access for A.Dot users is a concrete example of distribution-driven growth to watch.
- Regulatory actions around privacy and model training—these can temporarily remove apps from stores or force contractual changes, as seen with prior privacy pauses for certain entrants.
- Feature-driven surges: viral image/video features or device-level voice assistants can spike downloads, but retention depends on ongoing utility and governance.
- Enterprise procurement rhythms: corporate adoption of tenant-grounded copilots (Copilot, enterprise Claude, A.Dot Biz) will determine which assistants are present inside company workflows versus consumer pockets.
Final analysis — strengths, trade-offs, and a pragmatic verdict
The WiseApp·Retail November snapshot — showing ChatGPT, Perplexity, and A.Dot as the top three generative-AI apps by MAU in Korea — captures a mature market dynamic: a global generalist dominates by reach; research-first and telco-backed incumbents build meaningful, defensible niches through provenance and distribution. The numbers are directionally credible and consistent with earlier WiseApp reporting and industry signals like OpenAI’s Seoul presence, but they are not a substitute for vendor-level telemetry when making procurement choices. Notable strengths:
- ChatGPT’s product breadth and local investment support mass adoption and a rich plugin ecosystem.
- Perplexity’s evidence-first approach fills a distinct need for verifiable research and citation-aware workflows.
- A.Dot’s telco distribution, Korean-language focus, and agentic productivity features make it a compelling local alternative for users who value native integration and telco bundling.
Key risks:
- Privacy and regulatory action can rapidly reshape availability and trust — as DeepSeek’s experience showed.
- Hallucinations and data governance remain unresolved technical and policy issues; reliance on generative outputs without human review is dangerous for high-stakes decisions.
- Measurement differences across analytics vendors mean absolute MAU counts should be used with caution.
Bottom line: the WiseApp ranking is a useful snapshot and signals the state of play in November 2025 — ChatGPT remains the mass-market standard-bearer, Perplexity is the go-to for verifiable research, and A.Dot shows how local strategy and distribution can quickly convert into real user reach. Organizations and power users should pick the assistant that best matches their risk profile and workflows, enforce governance controls, and keep humans in the loop for anything consequential.
Acknowledgement of verification steps: the MAU figures and ranking above are reported by WiseApp·Retail and published in AsiaEconomy’s December 4, 2025 coverage of the WiseApp data. Earlier WiseApp snapshots and Korean press reporting confirm ChatGPT’s long-standing lead in Korea and the country’s unusually high subscriber counts for ChatGPT; OpenAI’s local investment and regulatory events such as app‑removals or privacy investigations (documented in the international press) corroborate both demand and the sensitivity of the market. Readers should treat third‑party analytics numbers as directional and consult vendor telemetry and contract terms before making operational decisions.
Source: 아시아경제
https://cm.asiae.co.kr/en/article/2025120408351091826/