AI Email Statistics 2026: How AI Changed Your Inbox

As of 2026, only about 13% of global email traffic is actually written by a human, and 51% of spam is now generated by AI. On the sending side, 79% of email senders use or plan to use AI in their programs. On the receiving side, the best documented result is that AI assistant users read about 11% fewer emails. The numbers below tell one story: AI has been overwhelmingly deployed to send more email, not to help you finish it.

AI email statistics measure two different arms races: how much of the mail entering your inbox was machine-written, and how many people use machines to get through it. Almost everyone aggregates the first race. Almost nobody aggregates the second. This page does both.

How many emails are written by AI?

The honest answer starts with a distinction most roundups skip: automated and AI-written are not the same thing.

On automation, the number is stark. Research from Hostinger, shared with TechRadar Pro in January 2026, found that only 13% of global email traffic is human-written — the other 87% comes from automated systems: marketing platforms, transactional senders, notification pipelines, and bots. Email stopped being a person-to-person medium a long time ago. AI just made the machines more fluent.

On generative AI specifically, the best dataset comes from researchers at Columbia University and the University of Chicago working with Barracuda. Analyzing unsolicited and malicious emails from February 2022 to April 2025 — neatly spanning the release of ChatGPT — they found that 51% of spam emails were AI-generated by April 2025. The majority of junk mail is now machine-authored, and the share climbed steadily after November 2022.

The same study found a slower creep in targeted attacks: 14% of business email compromise (BEC) attempts were AI-generated over the same period. Precision attacks still get human attention. Bulk annoyance gets the language model.

One caveat we will state plainly: nobody can measure "written by AI" perfectly. These figures rest on detection models, and detection is an arms race of its own. Treat the direction as solid and the decimals as soft.

How many emails are sent per day — and is AI growing the pile?

Email volume was growing before generative AI and kept growing after it. Industry trackers put worldwide volume at 361.6 billion emails per day in 2024, 376.4 billion in 2025, and a projected 392.5 billion in 2026, heading toward 408.2 billion by 2027.

Two details make that pile look worse up close. The Hostinger data suggests only 44% of sent email actually reaches an inbox — the rest is filtered or blocked en route. And of the blocked mail, 34% is flagged as phishing, scams, malware, or botnet traffic. The system is running hard just to keep the worst of it away from you.

What lands anyway is the subject of our email overload statistics — this page is about who is writing it.

How many email senders use AI?

Adoption on the sending side moved fast. In Mailjet by Sinch's Inbox Insights survey of over 3,200 senders, just 13.7% of email senders planned to adopt AI-powered tools in 2023. Three years later, the Sinch Mailgun Email Impact Report — a survey of 1,200+ senders plus analysis of 400+ billion emails sent through Mailgun infrastructure in 2025 — found 79% of senders use or plan to use AI in their email programs.

Year	Senders using or planning AI	Source
2023	13.7%	Mailjet by Sinch, Inbox Insights
2026	79%	Sinch Mailgun Email Impact Report

What are they using it for? Mostly writing:

AI use case among senders	Share	Source
Copy generation	41%	Sinch Mailgun, 2026
Content personalization	36%	Sinch Mailgun, 2026
Dynamic content	29%	Sinch Mailgun, 2026
Send-time optimization	27%	Sinch Mailgun, 2026
Data analysis	27%	Sinch Mailgun, 2026

Does it work for them? Modestly. 54% of senders actively using AI report moderate or significant year-over-year improvements — but so do 37% of senders who use no AI at all, and 23% of AI users report no meaningful benefit. The edge is real but thin. Which means the equilibrium is predictable: when everyone's copy is machine-polished, nobody's copy stands out — there is simply more of it, better targeted, in your inbox.

How many people use AI to handle the email they receive?

Now the other side of the ledger, the one nobody aggregates.

General workplace AI adoption looks impressive at first glance. Microsoft's Work Trend Index reported that 75% of knowledge workers use generative AI at work, with 46% of those users having started within the prior six months. Gallup's more conservative panel found 40% of US employees use AI at work in any form in 2025, up from 21% in 2023 — but only 19% use it a few times a week or more, and just 8% daily.

For email specifically, the receiving-side gains are measurable and small. Microsoft's Copilot research found users read 11% fewer individual emails and spent 4% less time in email, with some customers reporting 25–45% less time reading email. Real, but set it against the context from the same report: 85% of emails are read in under 15 seconds, and the typical person reads about 4 emails for every 1 they send. You are structurally on the losing side of that ratio.

None of this is new, incidentally. Google reported back in May 2017 that Smart Reply was already driving 12% of replies in Inbox on mobile. Machine-suggested replies are nearly a decade old. What changed since is the sophistication of the sending side, not the leverage of the receiving side. For how the time actually breaks down, see our email productivity statistics.

Why is the AI email arms race so lopsided?

Put the two sides in one table and the asymmetry is hard to miss:

Side	Metric	Number
Sending	Senders using or planning AI	79%
Sending	Spam that is AI-generated	51%
Sending	Emails sent daily, 2026 (projected)	392.5 billion
Receiving	Fewer emails read by Copilot users	11%
Receiving	US employees using AI daily, all tasks	8%

The economics explain it. Sending email is a revenue activity, so AI budget flows there; reading email is a cost center that belongs to you personally. A sender adopts AI once and the output lands in millions of inboxes. A reader adopts AI once and saves only their own minutes. The leverage compounds on exactly one side — which is how you get a decade of "AI will fix your inbox" promises alongside a volume line that only goes up. We wrote up what that does to the humans on the receiving end in the State of Inbox Overwhelm 2026.

Our own contribution to the receiving side is deliberately un-generative. We build Flick (flicked.email), a swipe-to-triage email client with a finite deck, and we maintain the Exit Gap Index — a crawler-seeded dataset that grades real senders A–F on whether they actually honor unsubscribe requests. No model wrote those grades; they are observed sender behavior. In an inbox increasingly written by machines, the scarce data is not more text. It is whether the sender respects a "no."

FAQ

What percentage of emails are written by AI in 2026?

About 51% of spam is AI-generated, and only about 13% of all email traffic is written by a human at all. The spam figure comes from Barracuda's research with Columbia University and the University of Chicago covering February 2022 to April 2025; the traffic figure comes from Hostinger research reported by TechRadar Pro. Note the distinction: "automated" includes ordinary marketing and transactional pipelines, while "AI-generated" means a model produced the text.

How many emails are sent per day in 2026?

Roughly 392.5 billion emails per day are projected for 2026, up from 376.4 billion in 2025. The volume line has risen every year on record and is projected to reach 408.2 billion by 2027. Generative AI did not start this growth, but it lowered the cost of producing each additional message to near zero.

Do AI email assistants actually save time?

Yes, but modestly: Copilot users read about 11% fewer emails and spend about 4% less time in email. Some organizations report 25–45% less reading time, so results vary with how deeply the tools are used. Against a sending side where 79% of senders use or plan to use AI, single-digit savings are unlikely to shrink your inbox on their own.

Is AI making spam worse?

AI made spam cheaper to produce and harder to spot by bad grammar alone — 51% of spam was AI-generated by April 2025, up steadily since ChatGPT launched. The same research found AI moving slower into targeted scams, with 14% of BEC attacks AI-generated. Filters are holding the line for now: Hostinger's data suggests only 44% of sent email reaches an inbox.

How many people use AI to write their own emails?

Precise "AI-written personal email" numbers do not exist yet; the closest proxies are workplace adoption figures. 40% of US employees use AI at work in some form, and 75% of knowledge workers report using generative AI, which includes drafting messages. Machine-assisted replies predate the current wave: Google's Smart Reply drove 12% of mobile replies in Inbox as far back as 2017.