ChatGPT Usage Statistics UK 2026

Statistic 11 • Generative AI

54% of UK Adults Now Say They Use AI Tools

Ofcom’s 2026 Adults’ Media Use and Attitudes research found that 54% of UK adults said they use AI tools such as ChatGPT, Microsoft Copilot or Google Gemini.

Statistic 12 • Generative AI

79% of UK 16–24-Year-Olds Use AI Tools

AI adoption rises sharply among younger adults. Approximately 79% of UK adults aged 16–24 reported using AI tools in Ofcom’s 2026 research.

Statistic 13 • Generative AI

74% of UK 25–34-Year-Olds Use AI Tools

Adoption is similarly high among the next age group, with approximately 74% of UK adults aged 25–34 saying they use AI tools.

Statistic 14 • CGO Media Calculation

AI Adoption Among 16–24s Is 25 Percentage Points Above the UK Adult Average

Comparing the reported 79% adoption rate among 16–24-year-olds with the 54% adult average produces a difference of 25 percentage points.

Statistic 15 • CGO Media Calculation

AI Adoption Among 25–34s Is 20 Percentage Points Above the Adult Average

The 74% adoption rate among 25–34-year-olds is 20 percentage points higher than the 54% figure reported across UK adults overall.

Statistic 16 • Generative AI

78% of UK 16–24-Year-Old Internet Users Had Already Used Generative AI in 2024

Ofcom’s June 2024 online research found that 78% of UK internet users aged 16–24 had used a generative-AI tool during the previous year, showing that younger adults were already early adopters before the wider acceleration measured in 2025 and 2026.

Statistic 17 • Generative AI

Only 14% of UK Internet Users Aged 65+ Had Used Generative AI in 2024

At the other end of the age distribution, only 14% of UK internet users aged 65 and over reported using generative AI in Ofcom’s June 2024 research.

Statistic 18 • CGO Media Calculation

Generative-AI Adoption Among 16–24s Was About 5.6 Times the Rate Among People Aged 65+

Using Ofcom’s 2024 age-group figures, the 78% adoption rate among 16–24-year-olds was approximately 5.6 times the 14% rate measured among internet users aged 65 and over.

Statistic 19 • ChatGPT Specific

27% of UK Online 18–24-Year-Olds Visited ChatGPT in May 2024

Ipsos iris audience measurement reported by Ofcom found that ChatGPT reached 27% of UK online adults aged 18–24 in May 2024, equivalent to approximately 1.4 million people in this age group.

Statistic 20 • ChatGPT Specific

62% of UK Adult ChatGPT Visitors Were Men in May 2024

Ofcom reported that 62% of ChatGPT’s measured UK adult visitors in May 2024 were men, representing approximately 2.6 million male visitors within the Ipsos iris audience dataset.

Central Finding

The UK AI audience remains strongly influenced by age. Younger adults adopted generative AI first and at much higher rates, making people under 35 particularly important for understanding how AI-assisted discovery may develop across the wider population.

AI Is Also Becoming More Personal

Ofcom’s 2026 research found that 12% of AI users used AI tools for conversational purposes. Among AI users aged 25–34, this increased to 19%.

These figures are not counted as ChatGPT-specific statistics because Ofcom’s question covered AI tools generally. They nevertheless illustrate how generative AI is expanding beyond straightforward productivity and information retrieval into conversational and advisory behaviour.

The Commercial Implication

The question for organisations is not simply “How many people use ChatGPT?” It is also which parts of their customer base are already using it heavily. Age, occupation, education and digital behaviour can materially change how quickly AI-assisted discovery becomes commercially important within a particular sector.

Primary Evidence Sources — Statistics 11–20

Ofcom — Adults’ Media Use and Attitudes 2026
Used for the 54% overall UK AI-tool adoption figure and the 79% and 74% adoption rates among 16–24 and 25–34-year-olds respectively.

Ofcom — Online Nation 2024
Used for the detailed 2024 age and gender breakdown of generative-AI adoption and ChatGPT-specific audience demographics.

Ipsos iris Online Audience Measurement Service
As reported by Ofcom, used for ChatGPT’s May 2024 reach among UK online 18–24-year-olds and the gender composition of its measured adult audience.

Statistic 21 • Global ChatGPT Research

Around 49% of Consumer ChatGPT Messages Fall Into the “Asking” Category

OpenAI’s large-scale study of consumer ChatGPT usage found that approximately 49% of messages were classified as “Asking” by mid-2025. These conversations seek information, guidance or advice that can help the user make a decision or understand something.

Statistic 22 • Global ChatGPT Research

Around 40% of Consumer ChatGPT Messages Are “Doing” Tasks

Approximately 40% of consumer ChatGPT messages were classified as “Doing” — requests where users ask ChatGPT to perform or assist with a task rather than primarily provide information or advice.

Statistic 23 • Global ChatGPT Research

Around 11% of Messages Are Classified as “Expressing”

The remaining approximately 11% of consumer ChatGPT messages were classified as “Expressing”, covering interactions that are neither primarily requests for information nor instructions to perform a practical task.

Statistic 24 • CGO Media Calculation

Approximately 89% of Consumer ChatGPT Messages Involve Asking or Doing

Combining the approximately 49% “Asking” share with the 40% “Doing” share indicates that roughly 89% of consumer ChatGPT messages fall into these two broad task-oriented categories.

Statistic 25 • Global ChatGPT Research

Practical Guidance Accounts for Around 29% of Consumer ChatGPT Conversations

OpenAI’s analysis found that Practical Guidance represented approximately 29% of consumer ChatGPT conversations, making it one of the largest identifiable usage categories.

Statistic 26 • Global ChatGPT Research

Seeking Information Represents Around 24% of Consumer ChatGPT Conversations

Approximately 24% of consumer ChatGPT conversations were classified as Seeking Information, demonstrating the scale at which ChatGPT is being used for information discovery.

Statistic 27 • Global ChatGPT Research

Writing Represents Around 24% of Consumer ChatGPT Conversations

Writing-related activity also accounted for approximately 24% of consumer ChatGPT conversations, placing writing alongside information seeking as one of the service’s dominant use cases.

Statistic 28 • CGO Media Calculation

Three Major Use Categories Account for Around 77% of Consumer Conversations

Practical Guidance (29%), Seeking Information (24%) and Writing (24%) together account for approximately 77% of consumer ChatGPT conversations.

Statistic 29 • Global ChatGPT Research

About 70% of Consumer ChatGPT Usage Is Unrelated to Work

OpenAI’s consumer-usage research found that approximately 70% of ChatGPT usage was non-work related by mid-2025, illustrating the extent to which the service has expanded into everyday personal activity.

Statistic 30 • Global ChatGPT Research

About 30% of Consumer ChatGPT Usage Is Work Related

Correspondingly, approximately 30% of consumer ChatGPT usage was classified as work related. Work therefore remains an important use case, but it represents a minority of total consumer usage.

Central Finding

ChatGPT is not primarily a writing machine. Information seeking, practical guidance and decision support form a substantial part of how consumers use the platform.

Question
→
Conversation
→
Clarification
→
Research
→
Comparison
→
Decision

Why This Matters for GEO and AI Search

If users were primarily using ChatGPT to generate text, the commercial relevance of AI visibility would be relatively narrow.

Instead, the data shows substantial usage for information seeking, practical guidance, advice and decision support. These activities can directly intersect with the customer journey.

For organisations, the visibility question therefore becomes: when users ask ChatGPT about subjects, problems, products or services connected to our market, does the AI system understand, retrieve, cite or recommend us?

Primary Evidence Source — Statistics 21–30

OpenAI / NBER — How People Use ChatGPT
Large-scale privacy-preserving analysis of consumer ChatGPT usage. Used for the Asking, Doing and Expressing classifications; Practical Guidance, Seeking Information and Writing categories; and the division between work and non-work activity.

Geographic scope: International ChatGPT consumer usage. These figures provide behavioural context and are not presented as UK-specific estimates.

Statistic 31 • Global ChatGPT Research

Approximately 30% of Consumer ChatGPT Usage Is Work Related

By mid-2025, approximately 30% of consumer ChatGPT usage was classified as work related. Professional activity therefore remains a substantial component of ChatGPT usage even as personal use has expanded more rapidly.

Statistic 32 • Global ChatGPT Research

Writing Accounts for Approximately 40% of Work-Related ChatGPT Messages

Within work-related conversations, writing represented approximately 40% of messages, making it the largest single category of professional ChatGPT activity in the research.

Statistic 33 • Global ChatGPT Research

About Two-Thirds of Work-Related Writing Requests Modify Existing Text

Approximately two-thirds of writing-related work messages involved modifying text supplied by the user rather than asking ChatGPT to create entirely new material.

Statistic 34 • CGO Media Calculation

Roughly 27% of Work-Related ChatGPT Messages Could Involve Editing Existing Text

If writing accounts for approximately 40% of work-related messages and roughly two-thirds of those writing interactions involve modifying existing text, the two reported shares imply that approximately 27% of work-related messages could fall into this editing or transformation category.

Statistic 35 • Global ChatGPT Research

About 56% of Work-Related ChatGPT Messages Are Classified as “Doing”

OpenAI’s analysis found that approximately 56% of work-related ChatGPT messages were classified as Doing, meaning the user was asking the system to perform or help complete a task.

Statistic 36 • Global ChatGPT Research

Around 35% of Work-Related Messages Are Classified as “Asking”

Approximately 35% of work-related ChatGPT messages were classified as Asking, where the user seeks information or advice that can support understanding, judgement or a professional decision.

Statistic 37 • CGO Media Calculation

Approximately 91% of Work-Related Messages Are Either Asking or Doing

Combining the reported 56% Doing share and 35% Asking share indicates that approximately 91% of work-related ChatGPT messages fall within these two broad categories.

Statistic 38 • Global ChatGPT Research

Computer Programming Represents About 4% of All Consumer ChatGPT Messages

Despite ChatGPT’s prominent association with software development, computer programming accounted for approximately 4% of consumer messages in the usage research.

Statistic 39 • Global ChatGPT Research

About 10% of Work-Related Messages Involve Technical Help

Technical Help accounted for approximately 10% of work-related ChatGPT messages, encompassing programming, mathematical calculation, data analysis and related technical activities.

Statistic 40 • Global ChatGPT Research

Management and Business Occupations Show Particularly High Work-Related ChatGPT Use

OpenAI’s occupational analysis found that approximately 40% of messages from users in management and business occupations were work related, demonstrating how professional ChatGPT usage varies substantially by occupation.

Central Finding

Professional ChatGPT usage is dominated by knowledge work: writing, editing, obtaining information, seeking guidance, technical assistance and completing cognitive tasks.

The AI-Assisted Knowledge-Work Cycle

Research
→
Draft
→
Analyse
→
Edit
→
Critique
→
Refine
→
Decision

What This Means for Organisations

The workplace impact of ChatGPT cannot be measured simply by asking whether a job can be automated.

AI can sit inside existing workflows as a research assistant, editor, analyst, translator, critic, tutor, technical assistant or decision-support system.

For many knowledge workers, the emerging model is therefore not Human versus AI. It is increasingly Human + AI versus the same workflow performed without AI assistance.

Primary Evidence Source — Statistics 31–40

OpenAI / NBER — How People Use ChatGPT
Large-scale privacy-preserving research into consumer ChatGPT usage. Used for work versus non-work activity, workplace writing, Asking and Doing behaviour, technical assistance, programming and occupational differences.

Geographic scope: International consumer ChatGPT usage. Findings in this section are behavioural evidence and are not presented as UK workforce adoption rates.

Statistic 41 • Global ChatGPT Research

Around 24% of Consumer ChatGPT Conversations Involve Seeking Information

OpenAI’s consumer-usage research found that approximately 24% of ChatGPT conversations were classified as Seeking Information, making information retrieval one of the platform’s three largest use categories.

Statistic 42 • Global ChatGPT Research

Practical Guidance Accounts for Around 29% of Consumer ChatGPT Conversations

Approximately 29% of consumer ChatGPT conversations were classified as Practical Guidance, covering activities such as advice, tutoring, ideas and assistance with real-world problems.

Statistic 43 • CGO Media Calculation

Information Seeking and Practical Guidance Together Represent Around 53% of Conversations

Combining the reported 24% Seeking Information share and 29% Practical Guidance share produces approximately 53% of consumer ChatGPT conversations.

Statistic 44 • Global ChatGPT Research

Approximately 49% of Consumer Messages Are Classified as “Asking”

By mid-2025, approximately 49% of consumer ChatGPT messages were classified as Asking — interactions in which users seek information, advice or guidance rather than primarily instructing the system to produce an output.

Statistic 45 • Independent AI Search Research

Around 65% of Measured ChatGPT Activity Has Been Classified as Search or Search-Like

Independent analysis by Ahrefs classified approximately 65% of measured ChatGPT activity as search or search-like intent. This broader definition includes informational behaviour that may overlap with activities historically performed through search engines.

Statistic 46 • Independent AI Search Research

ChatGPT Used Web Search on Around 34.5% of Measured Queries

Research examining ChatGPT’s use of external search found that by early 2026 the system was using web search on approximately 34.5% of measured queries, demonstrating the increasing integration of live web retrieval into the ChatGPT experience.

Statistic 47 • UK ChatGPT Research

13% of UK Adults Use ChatGPT to Keep Up With News

Ofcom’s 2026 news research found that 13% of UK adults use ChatGPT to keep up with news, making ChatGPT an increasingly relevant gateway to current information as well as a general-purpose AI assistant.

Statistic 48 • UK Generative AI Research

Around 20% of UK Adults Use AI Tools to Keep Up With News

Across AI services more broadly, approximately one in five UK adults reported using AI tools to keep up with news in Ofcom’s 2026 research.

Statistic 49 • CGO Media Calculation

ChatGPT Alone Accounts for a Large Share of Reported UK AI News Use

Comparing ChatGPT’s 13% UK adult news-use figure with the approximately 20% using AI tools generally for news indicates that ChatGPT’s reported reach is equivalent to around 65% of the broader AI-news-use rate. This is a comparison of population rates, not a claim that 65% of AI-news users use ChatGPT.

Statistic 50 • UK AI Information Behaviour

60% of UK Adults Report Seeing AI-Generated Search Summaries

Ofcom’s 2026 research found that approximately 60% of UK adults had encountered AI-generated summaries within search results. This is not a ChatGPT-specific statistic, but it demonstrates how conversational and generative information delivery is spreading beyond dedicated chatbot interfaces.

Central Finding

ChatGPT is becoming part of the information-discovery environment. Users are not only asking it to create things; they are using it to find information, understand subjects, obtain guidance, research decisions and increasingly access current web information.

From Keyword Search to Conversational Discovery

Question
→
Context
→
Research
→
Follow-Up
→
Comparison
→
Recommendation

Why Search-Like ChatGPT Usage Matters for GEO

Traditional SEO asks whether a page can rank prominently when someone searches for a relevant query.

Generative Engine Optimisation introduces additional questions: can an organisation’s information be retrieved by an AI system, recognised as relevant evidence, selected as a source, cited in an answer or incorporated into a recommendation?

As information-seeking behaviour spreads across search engines and conversational AI, visibility increasingly needs to be measured across the entire discovery ecosystem rather than through rankings alone.

AI Search Visibility Model

Entity Understanding
→
Retrieval
→
Source Selection
→
Citation
→
Brand Mention
→
Recommendation

Primary Evidence Sources — Statistics 41–50

OpenAI / NBER — How People Use ChatGPT
Used for Seeking Information, Practical Guidance and Asking behaviour within consumer ChatGPT usage.

Ahrefs — ChatGPT Search and Search-Like Behaviour Research
Used for independent estimates of the proportion of ChatGPT activity classified as search or search-like.

Independent ChatGPT Web-Search Analysis
Used for the measured proportion of queries in which ChatGPT performed external web search.

Ofcom — UK News Consumption / AI News Research 2026
Used for UK adoption of ChatGPT and AI tools for news and wider evidence concerning exposure to AI-generated information within search.

Statistic 51 • Independent Traffic Research

ChatGPT Referral Traffic Increased by Approximately 206% During 2025

Semrush clickstream research found that measured outbound referral traffic from ChatGPT increased by approximately 206% during 2025, demonstrating rapid growth in the number of journeys moving from ChatGPT to external websites.

Statistic 52 • Independent Traffic Research

More Than 30% of Measured ChatGPT Referral Traffic Went to Just 10 Domains

The same research found that more than 30% of measured ChatGPT referral traffic flowed to only ten domains, showing that outbound traffic was highly concentrated among a relatively small number of destinations.

Statistic 53 • Independent Traffic Research

More Than 20% of Measured ChatGPT Referral Traffic Went to Google

Google alone received more than 20% of measured outbound ChatGPT referral traffic in the Semrush clickstream analysis.

Statistic 54 • CGO Media Interpretation

At Least Two-Thirds of Traffic Going to the Top 10 Destinations Was Accounted for by Google

Because Google received more than 20% of all measured referrals while the ten largest destinations collectively received more than 30%, the reported lower bounds indicate that Google represented at least two-thirds of the minimum traffic represented by those two thresholds. This is a comparison of published lower bounds, not an estimate of Google’s exact share within the top ten.

Statistic 55 • Independent Website Traffic Research

ChatGPT Accounted for Approximately 0.21% of Measured Website Traffic

An Ahrefs analysis of approximately 76,000 websites estimated that ChatGPT accounted for roughly 0.21% of measured website traffic within the dataset.

Statistic 56 • Independent Website Traffic Research

Google Accounted for Nearly 40% of Measured Website Traffic

Within the same Ahrefs dataset, Google accounted for nearly 40% of measured website traffic, illustrating the enormous difference between current Google referral scale and measurable ChatGPT referrals.

Statistic 57 • CGO Media Calculation

Google Sent Roughly 190 Times More Measured Website Traffic Than ChatGPT

Comparing approximately 40% of measured traffic from Google with approximately 0.21% from ChatGPT gives a ratio of roughly 190 to one.

Statistic 58 • Independent Website Traffic Research

AI Platforms Still Represent a Small Share of Direct Website Referral Traffic

Across the approximately 76,000-site dataset, measured AI-platform referrals represented less than 1% of total website traffic, despite rapid growth in the use of AI assistants.

Statistic 59 • Independent Traffic Research

ChatGPT’s Referral Growth Rate Was More Than Three Times Its Starting Level

A 206% increase means measured ChatGPT referral traffic reached approximately 3.06 times its starting level over the period covered by the research.

Statistic 60 • UK ChatGPT Context

UK Users Generated Approximately 1.8 Billion ChatGPT Visits in Eight Months

Ofcom reported approximately 1.8 billion UK visits to ChatGPT during the first eight months of 2025. Although this statistic was introduced earlier in the report as an audience-growth measure, its traffic significance is important here: enormous usage inside ChatGPT can coexist with a comparatively small direct referral share to external websites.

Important Measurement Principle

ChatGPT usage, ChatGPT referrals and ChatGPT influence are three different measurements. A platform can influence a consumer decision without generating a direct click to the organisation being considered.

The Emerging Cross-Platform Search Journey

ChatGPT
→
Brand Discovery
→
Google
→
Website
→
Reviews
→
ChatGPT
→
Decision

Example:
User asks ChatGPT: “Which enterprise SEO consultancies specialise in AI search?”

ChatGPT mentions several organisations.

The user later searches one company name in Google and visits its website.

The measurable website visit may be attributed to Google even though the original brand discovery occurred in ChatGPT.

CGO Media AI Discovery Measurement Model

Source Visibility
→
Citation Visibility
→
Brand Visibility
→
Recommendation Visibility
→
Referral Traffic
→
Commercial Outcome

The Strategic Point

The evidence does not justify abandoning traditional SEO for ChatGPT optimisation. It supports a broader strategy in which organisations maintain Google visibility while also building the entity, authority, content and citation signals required for AI-mediated discovery.

Primary Evidence Sources — Statistics 51–60

Semrush — ChatGPT Search & Referral Behaviour Research
Clickstream analysis used for ChatGPT outbound referral growth, traffic concentration and the proportion of measured outbound referrals reaching Google.

Ahrefs — ChatGPT vs Google Website Traffic Research
Analysis of approximately 76,000 websites used to compare measured website traffic originating from ChatGPT, AI platforms and Google.

Ofcom — The Era of Answer Engines
Used as UK context for the approximately 1.8 billion ChatGPT visits recorded during the first eight months of 2025.

Statistic 61 • Global ChatGPT

ChatGPT Reached 1 Million Registered Users Within Days of Launch

ChatGPT was released publicly on 30 November 2022 as a research preview. By 5 December 2022 it had exceeded one million registered users.

Statistic 62 • Global ChatGPT

ChatGPT Exceeded 100 Million Logged-In Weekly Active Users After One Year

OpenAI’s economic research reports that ChatGPT had reached more than 100 million logged-in weekly active users after approximately one year.

Statistic 63 • Global ChatGPT

Weekly Active Users Approached 350 Million After Two Years

By approximately two years after launch, ChatGPT had reached almost 350 million logged-in weekly active users on consumer plans.

Statistic 64 • Global ChatGPT

ChatGPT Surpassed 700 Million Weekly Active Users by July 2025

By the end of July 2025, ChatGPT had more than 700 million total weekly active users, according to OpenAI’s large-scale usage research.

Statistic 65 • Global ChatGPT

ChatGPT Reached More Than 800 Million Weekly Users During 2025

OpenAI subsequently reported that ChatGPT was serving more than 800 million users every week, demonstrating that growth continued after the July 2025 research snapshot.

Statistic 66 • Global ChatGPT

ChatGPT Reached More Than 900 Million Weekly Active Users in 2026

OpenAI reported in 2026 that ChatGPT had grown to more than 900 million weekly active users, continuing the platform’s expansion beyond the 800-million level reported during 2025.

Statistic 67 • Global ChatGPT

Users Were Sending More Than 2.5 Billion Messages Per Day by July 2025

OpenAI reported that users were sending more than 2.5 billion messages per day through its platform by July 2025.

Statistic 68 • Global ChatGPT

ChatGPT Was Handling Approximately 18 Billion Messages Per Week

OpenAI’s usage paper reports that by July 2025 approximately 18 billion messages were being sent each week across ChatGPT.

Statistic 69 • Global ChatGPT

ChatGPT Was Processing Around 29,000 User Messages Every Second

The daily message volume equated to approximately 29,000 messages every second by July 2025, illustrating the operational scale of consumer ChatGPT usage.

Statistic 70 • Global ChatGPT

Message Volume Increased More Than Fivefold in One Year

OpenAI’s usage research found that between July 2024 and July 2025 the total number of ChatGPT messages sent by users increased by more than five times.

Central Finding

ChatGPT’s expansion is not only an audience-growth story. Message volume has risen dramatically as the user base has expanded, indicating increasing depth as well as breadth of adoption.

ChatGPT Adoption Has Two Dimensions

More Users
+
More Messages
+
More Tasks
+
More Frequent Use
=
Deeper Adoption

Why Global Scale Matters to UK Organisations

Global adoption matters even to organisations operating primarily in the UK because AI platforms are developing at international scale. The underlying systems, interfaces and user behaviours are not being shaped by the UK market alone.

For companies serving international customers, the effect is even more direct. A website can increasingly be encountered not only through country-specific Google results but through AI systems used across many markets.

ChatGPT’s scale makes AI visibility an international discovery issue rather than a niche technology experiment.

Primary Evidence Sources — Statistics 61–70

OpenAI / NBER — How People Use ChatGPT
Used for ChatGPT’s early growth, 100 million and approximately 350 million logged-in weekly-active-user milestones, the 700-million weekly-user level, 18-billion weekly message volume, message growth and the approximately 29,000-messages-per-second figure.

OpenAI — State of Enterprise AI / 2025 Adoption Updates
Used for the subsequent milestone of more than 800 million weekly ChatGPT users.

OpenAI — Scaling AI for Everyone, 2026
Used for the updated figure of more than 900 million weekly active ChatGPT users.

OpenAI — ChatGPT Adoption Research, 2026
Used for supporting evidence showing that users sent 50% more daily messages after six months and doubled the number of task types they had tried.

Statistic 71 • Global ChatGPT Research

Around 80% of Early ChatGPT Users Had Typically Masculine First Names

OpenAI’s privacy-preserving demographic analysis found that among users whose first names could be classified, approximately 80% of active users in the early months after launch had names typically associated with men.

Statistic 72 • CGO Media Calculation

Only Around 20% of the Early Classifiable User Base Had Typically Feminine First Names

The approximately 80% masculine-name share implies that only around 20% of the classifiable early user base had names typically associated with women.

Statistic 73 • Global ChatGPT Research

The Gender Gap Had Almost Completely Closed by Mid-2025

By mid-2025, OpenAI’s demographic analysis showed that the earlier gender imbalance had narrowed dramatically and was close to parity among users whose first names could be classified.

Statistic 74 • Global ChatGPT Research

Typically Feminine Names Slightly Exceeded Half of the Classifiable User Base by Mid-2025

By mid-2025, users with first names typically associated with women represented slightly more than half of the classifiable consumer user base, reversing the strong male skew seen shortly after launch.

Statistic 75 • Global ChatGPT Research

Adults Under 26 Accounted for Nearly Half of Adult ChatGPT Messages

OpenAI’s age analysis found that users aged 18–25 accounted for nearly half of all messages sent by adult users in the studied consumer ChatGPT population.

Statistic 76 • Global ChatGPT Research

Younger Adults Use ChatGPT More Heavily for Non-Work Purposes

Among adult users aged 18–25, only approximately 23% of messages were work related, meaning the majority of their ChatGPT activity occurred outside work.

Statistic 77 • CGO Media Calculation

Approximately 77% of Messages From 18–25-Year-Old Adults Were Non-Work Related

If approximately 23% of messages from adult users aged 18–25 were work related, the corresponding non-work share was approximately 77%.

Statistic 78 • Global ChatGPT Research

Work-Related Usage Rises With Age Through Much of the Adult Population

OpenAI’s research found that the proportion of messages classified as work related generally increased with user age through the main working-age groups, compared with the approximately 23% share among 18–25-year-olds.

Statistic 79 • Global ChatGPT Research

Users in Lower-Income Countries Have Been Adopting ChatGPT Faster

OpenAI’s geographic analysis found that ChatGPT adoption growth was faster in lower-income countries than in the highest-income countries, indicating that the user base was becoming more geographically diverse as the platform expanded.

Statistic 80 • Global ChatGPT Research

ChatGPT Was Already Used by Roughly 10% of the World’s Adult Population by Mid-2025

The more than 700 million weekly active users reported by July 2025 represented approximately 10% of the global adult population, according to OpenAI’s usage research.

Central Finding

ChatGPT’s demographic profile has broadened as adoption has grown. What began as a heavily male-skewed early-adopter audience developed into a much more balanced mass-market user base, while younger adults continue to account for a particularly large share of usage.

Important: How OpenAI Estimated Gender

The gender analysis did not identify or verify each user’s gender.

OpenAI’s research classified first names according to whether those names are more commonly associated with men or women. The results should therefore be described as typically masculine or typically feminine first-name classifications, rather than as direct measurements of users’ gender identity.

Learning
→
Personal Use
→
Professional Use
→
Decision Support
→
Everyday AI Utility

Why Demographics Matter for AI Visibility

The commercial importance of ChatGPT will not arrive at the same speed in every market.

Businesses serving younger audiences may encounter AI-mediated research and discovery earlier and more frequently. Organisations serving older populations may experience a slower transition, although mainstream adoption can narrow demographic gaps over time.

AI-search strategy should therefore consider who the organisation’s audience is, not merely the overall size of ChatGPT’s global user base.

Primary Evidence Source — Statistics 71–80

OpenAI / NBER — How People Use ChatGPT
Used for demographic changes in the consumer ChatGPT user base, first-name-based gender classification, age-related usage patterns, work versus non-work activity and geographic adoption trends.

Geographic scope: Global consumer ChatGPT usage. These findings provide international context and are not presented as UK demographic estimates.

Statistic 81 • Global ChatGPT Research

10.2% of Consumer ChatGPT Messages Are Tutoring or Teaching Requests

OpenAI's large-scale analysis found that 10.2% of all sampled consumer ChatGPT messages were classified as Tutoring or Teaching. Education is therefore a substantial identifiable use case within overall ChatGPT activity.

Statistic 82 • Global ChatGPT Research

36% of Practical Guidance Messages Involve Tutoring or Teaching

Within the broader Practical Guidance category, 36% of messages were classified as Tutoring or Teaching, making learning one of the largest components of ChatGPT's practical-guidance activity.

Statistic 83 • Global ChatGPT Research

8.5% of Consumer ChatGPT Messages Are General How-To Advice

A further 8.5% of all consumer ChatGPT messages were classified as general how-to advice across a wide range of topics, demonstrating that informal learning extends beyond conversations explicitly categorised as tutoring.

Statistic 84 • Global ChatGPT Research

30% of Practical Guidance Messages Are General How-To Advice

General how-to advice represented approximately 30% of Practical Guidance messages in OpenAI's conversation analysis.

Statistic 85 • CGO Media Calculation

Tutoring and General How-To Advice Together Represent 18.7% of Messages

Combining the 10.2% Tutoring or Teaching share with the 8.5% general how-to share gives approximately 18.7% of all consumer ChatGPT messages across these two learning and guidance categories.

Statistic 86 • UK Children

37% of UK 8–14-Year-Olds Visited an AI Service in a Month

Ofcom's passive online measurement research found that 37% of UK children aged 8–14 visited an artificial-intelligence service during the measured month, equivalent to approximately 2.1 million children.

Statistic 87 • UK Children

50% of UK 13–14-Year-Olds Visited an AI Service in the Measured Month

AI-service reach increased sharply with age. Among UK children aged 13–14, Ofcom measured monthly AI-service reach at 50%.

Statistic 88 • UK Children

35% of UK 10–12-Year-Olds Visited an AI Service

Among children aged 10–12, 35% visited an AI service during Ofcom's measured month.

Statistic 89 • UK Children

27% of UK 8–9-Year-Olds Visited an AI Service

Even among the youngest age group in Ofcom's passive measurement study, 27% of UK 8–9-year-olds visited an AI service during the measured month.

Statistic 90 • UK ChatGPT

9% of UK 8–14-Year-Olds Visited ChatGPT in a Month

ChatGPT itself reached 9% of UK children aged 8–14 during Ofcom's measured month, equivalent to approximately 509,000 children. It was the second-highest-reaching standalone AI service in the study.

Central Finding

Learning is not a peripheral ChatGPT use case. Tutoring or teaching alone accounts for around one in ten consumer messages, while UK evidence shows that exposure to AI services is already substantial among school-age children.

The Conversational Learning Model

Question
→
Explanation
→
Follow-Up
→
Example
→
Practice
→
Feedback
→
Understanding

Why This Matters Beyond Education

Students and younger users who become comfortable asking AI systems questions may carry that behaviour into employment, purchasing, travel, finance and other areas of adult life.

The long-term significance is therefore not simply whether students use ChatGPT to complete coursework. It is whether conversational interaction becomes a normal way of finding information, learning about unfamiliar subjects and evaluating choices.

For organisations, today's educational AI behaviour may help shape tomorrow's mainstream search and discovery behaviour.

Primary Evidence Sources — Statistics 81–90

OpenAI / NBER — How People Use ChatGPT
Used for the 10.2% Tutoring or Teaching share, 36% of Practical Guidance, 8.5% general how-to advice and the 30% Practical Guidance how-to share.

Ofcom — Children's Passive Online Measurement
Used for monthly UK AI-service reach among 8–14-year-olds, the age breakdown and ChatGPT's 9% / approximately 509,000 reach among UK children aged 8–14.

Ofcom — Generative AI Adoption Research
Used as supporting evidence for study-related generative-AI use among UK internet users aged 16 and above.

Statistic 91 • OpenAI Signals 2026

Users Send 50% More Messages Per Day Six Months After Signing Up

OpenAI Signals found that six months after account creation, sampled ChatGPT users were sending approximately 50% more messages per day than they sent during their initial period of usage.

Statistic 92 • OpenAI Signals 2026

Users Double the Number of Distinct ChatGPT Tasks They Try Within Six Months

The same research found that after six months users had doubled the number of distinct task types they had tried with ChatGPT compared with their initial usage period.

Statistic 93 • OpenAI Signals Methodology

The Deepening-Adoption Study Uses a 0.1% Sample of ChatGPT Users

OpenAI’s 2026 analysis of usage depth and breadth was based on a 0.1% sample of users whose accounts were created between 15 October 2025 and 1 May 2026, with activity measured through 31 May 2026.

Statistic 94 • OpenAI Signals 2026

Users Under 35 Still Accounted for the Largest Share of ChatGPT Messages in Q1 2026

OpenAI reported that users under the age of 35 continued to account for the largest share of total consumer ChatGPT messages during the first quarter of 2026.

Statistic 95 • OpenAI Signals 2026

Users Over 35 Increased Their Share of ChatGPT Messages in Q1 2026

Although younger users remained the largest group, OpenAI found that users aged over 35 gained share of total messages during Q1 2026, providing evidence that adoption was broadening beyond younger early adopters.

Statistic 96 • OpenAI Signals 2026

Users With Typically Feminine Names Represent More Than Half of Classifiable Users

OpenAI’s Q1 2026 research found that users with names typically associated with women represented more than half of users for whom gender could be inferred from first names, continuing the demographic broadening identified during 2025.

Statistic 97 • OpenAI Signals 2026

Multimedia Accounts for 7.8% of Global ChatGPT Messages

OpenAI’s August 2026 country-level usage analysis found that multimedia accounted for 7.8% of ChatGPT messages globally, making it one of the fastest-growing areas of consumer usage.

Statistic 98 • OpenAI Signals 2026

Multimedia Exceeds 10% of Messages in Some Countries

Multimedia represented more than one in ten ChatGPT messages in countries including Brazil and Colombia, illustrating how usage is expanding beyond text-based questions and answers.

Statistic 99 • OpenAI Signals 2026

At Work, Users Are More Than Twice as Likely to Ask ChatGPT to “Do” Something

OpenAI’s 2026 country-level analysis found that in work contexts people were more than twice as likely to use ChatGPT to complete a task or create something as they were outside work.

Statistic 100 • UK AI Discovery 2026

21% of UK Adults Used an AI App or Service to Access News in the Past Month

Ofcom’s News Consumption 2026 research found that 21% of UK adults had used an AI app or service to access news during the previous month. ChatGPT was the most widely used individual AI tool for news, reaching 13% of adults.

100 Statistics Complete

ChatGPT Has Moved From Adoption to Embedded Usage

The central story emerging from these 100 statistics is no longer simply that more people are trying ChatGPT. Existing users are increasing the frequency, breadth and variety of their usage while the platform expands across demographics, countries and activities.

The ChatGPT Adoption Cycle

Experiment
→
Learn
→
Repeat
→
Expand Tasks
→
Integrate
→
Habitual Use

Text
→
Images
→
Documents
→
Data
→
Research
→
Multimodal Assistance

The 2026 Search & Discovery Ecosystem

Google Search
+
AI Overviews
+
ChatGPT
+
Other AI Assistants
+
Social Platforms
+
Publishers
+
Direct Websites

What the 100 Statistics Mean for UK Organisations

The evidence does not support the claim that ChatGPT has replaced Google. Google remains vastly larger as a measurable source of website traffic.

The evidence does support a different conclusion: ChatGPT has become a significant additional environment in which people ask questions, seek information, obtain advice, research choices and encounter organisations.

For businesses, institutions and publishers, search visibility in 2026 increasingly means being discoverable across both traditional search engines and AI-mediated information systems.

CGO Media Research

The Emerging AI Discovery Journey

User Need
→
Question
→
AI Retrieval
→
Source Selection
→
Answer
→
Comparison
→
Recommendation
→
Action

Primary Evidence Sources — Statistics 91–100

OpenAI Signals — How ChatGPT Adoption Has Expanded, June 2026
Used for the 50% increase in daily messages after six months, doubling of distinct tasks and the 0.1% longitudinal user sample.

OpenAI Signals — How ChatGPT Adoption Broadened in Early 2026
Used for age and inferred-gender changes in Q1 2026 and evidence that the consumer user base continued to broaden beyond early adopters.

OpenAI — From Asking to Doing: How the World Is Putting ChatGPT to Work, August 2026
Used for global multimedia usage and the difference between Doing behaviour in work and non-work contexts.

Ofcom — News Consumption in the UK 2026
Used for the 21% of UK adults accessing news through AI apps or services and the 13% using ChatGPT for news.

Finding 1

ChatGPT Has Achieved Mass-Market Scale

The global user base expanded from one million registered users shortly after launch to hundreds of millions of weekly active users. By 2026, OpenAI was reporting more than 900 million weekly active users.

In the UK, ChatGPT’s measured adult audience almost tripled between June 2024 and June 2025. This makes it increasingly difficult to treat conversational AI as a niche channel relevant only to technology-focused audiences.

Finding 2

Adoption Deepens After People Start Using ChatGPT

OpenAI’s 2026 longitudinal evidence is important because it shows that adoption is not simply a question of acquiring new users. Six months after joining, sampled users were sending approximately 50% more messages per day and had tried twice as many task types.

This suggests a progression from experimentation toward integration. People discover additional uses for ChatGPT as familiarity grows.

Finding 3

Information Seeking Is a Core ChatGPT Behaviour

Writing and content creation remain important uses, but the evidence does not support describing ChatGPT primarily as a writing tool. Seeking Information and Practical Guidance together account for a substantial proportion of consumer conversations.

Users ask questions, research unfamiliar subjects, seek advice, compare possibilities and request explanations. Those behaviours increasingly overlap with functions traditionally associated with search.

Finding 4

AI Discovery Does Not Require an Immediate Website Click

Measured ChatGPT referral traffic remains small compared with Google. That fact is important and should not be obscured by exaggerated claims about the death of traditional search.

However, referral traffic captures only part of AI influence. ChatGPT can introduce a brand, explain a product, compare providers or recommend an organisation without producing an immediate referral session. The resulting website visit may occur later through Google, direct navigation or another channel.

Finding 5

Search Is Becoming an Ecosystem Rather Than a Single Interface

Consumers can now move between Google, ChatGPT, AI-generated search summaries, social platforms, specialist publishers, review sites and company websites during the same research journey.

The strategic question for organisations is therefore becoming broader than “Where do we rank?” It increasingly includes “Where are we retrieved, cited, mentioned, compared and recommended?”

The strategic shift is not
“SEO or GEO”.

It is
“SEO + AI Search + GEO + Entity Authority + Citation Visibility”.

Need
→
AI Question
→
Retrieved Sources
→
AI Answer
→
Brand Discovery
→
Comparison
→
Website / Search
→
Decision

Rankings
+
Organic Traffic
+
AI Citations
+
Brand Mentions
+
Recommendation Visibility
+
AI Referrals
+
Commercial Outcomes

Final Conclusion

Search Visibility Is Becoming Discovery Visibility

ChatGPT has not made websites irrelevant. It has not eliminated Google. It has not removed the need for technical SEO, authoritative content or strong brands.

Instead, it has added another influential layer between a user’s question and the sources, organisations and products they eventually choose.

For organisations operating in 2026, visibility increasingly means being discoverable, understandable, retrievable, citable and recommendable wherever people ask questions.

Suggested Citation

CGO Media Research Team (2026), ChatGPT Usage Statistics UK 2026: 100 Statistics on Adoption, Search, Work & AI Behaviour, CGO Media.

Research should be interpreted alongside the methodology, source dates and geographic scope stated throughout this report. UK-specific evidence is distinguished from international ChatGPT research and CGO Media calculations.

Key Ofcom Research Used

Online Nation 2025
Used for UK ChatGPT audience reach, year-on-year growth and wider UK online-service usage.

The Era of Answer Engines
Used for evidence on the growth of ChatGPT visits and the changing relationship between conventional search and generative AI.

Generative AI Adoption Research
Used for age-related UK adoption and evidence concerning work, study and recreational use of generative AI.

Children’s Passive Online Measurement
Used for AI-service and ChatGPT reach among UK children aged 8–14.

News Consumption in the UK 2026
Used for UK adult use of AI services and ChatGPT as routes to news and current information.

How ChatGPT Adoption Has Expanded
Used for evidence that sampled users sent approximately 50% more messages per day six months after account creation and doubled the number of distinct task types they had tried.

2026 Demographic Adoption Research
Used for evidence that older users increased their share of messages and that the gender composition of classifiable users continued to broaden.

2026 Global Usage Research
Used for evidence concerning multimedia usage and differences between work and non-work ChatGPT behaviour.

For example:

“54% of UK adults use AI tools” is not equivalent to saying “54% of UK adults use ChatGPT”.

“37% of UK 8–14-year-olds visited an AI service” is not equivalent to saying 37% visited ChatGPT.

Where ChatGPT-specific measurements exist, they are identified separately.

Key Limitations

1. Different time periods. Studies were conducted at different points between 2024 and 2026. Rapid platform growth means values from different periods should not be treated as simultaneous measurements.

2. Different populations. Some evidence relates to UK adults, some to UK internet users, some to children and some to global ChatGPT users.

3. Different methodologies. Survey responses, passive audience measurement, platform telemetry and clickstream datasets measure different forms of behaviour.

4. ChatGPT is evolving. Product capabilities including web search, multimodal interaction and other tools can change usage behaviour over relatively short periods.

5. Referral data does not capture all influence. A ChatGPT recommendation can affect later behaviour even where the user does not immediately click through to a website.

6. Classification is study-specific. Categories such as Asking, Doing, Seeking Information and search-like behaviour depend on the methodology and definitions used by the original research.

These 100 statistics should not be interpreted as 100 independent measurements of the same phenomenon.

Together they describe different dimensions of a rapidly evolving ecosystem: audience reach, usage intensity, demographic adoption, work behaviour, education, information seeking, search-like activity, referrals and wider AI-assisted discovery.

The strongest conclusions are those supported by convergence across multiple independent evidence sources rather than by any single statistic in isolation.

Ofcom. Online Nation 2025.

Ofcom. The Era of Answer Engines.

Ofcom. Research on generative-AI adoption among UK internet users.

Ofcom. Children’s Passive Online Measurement.

Ofcom. News Consumption in the UK 2026.

OpenAI / NBER. How People Use ChatGPT.

OpenAI. How ChatGPT Adoption Has Expanded. 2026.

OpenAI. 2026 research on the demographic broadening of ChatGPT adoption.

OpenAI. 2026 global ChatGPT usage research covering work behaviour and multimedia activity.

Semrush. Research examining ChatGPT outbound referrals and clickstream behaviour.

Ahrefs. Research examining ChatGPT referral traffic, AI traffic and search-like behaviour.

CGO Media Research Standard

Evidence Before Commentary

CGO Media research distinguishes measured evidence from interpretation. Statistics are attributed to their underlying sources, calculations are identified and geographic limitations are stated where relevant.

The objective is not to exaggerate the displacement of traditional search, but to measure how search and discovery are changing as AI-mediated information systems become part of everyday behaviour.