What buyers actually say about AI-generated outreach, according to 2026 survey data

Two surveys published within weeks of each other in 2026, one aimed at B2B buyers specifically and one at general email consumers, land on a version of the same answer when you ask what people actually think about AI-generated outreach. It’s more specific — and more useful — than “buyers hate AI email.”

What Gartner found among B2B buyers

Gartner’s 2026 survey of 645 B2B buyers, fielded in late 2025 and presented at Gartner’s CSO & Sales Leader Conference in May 2026, found a pattern that looks contradictory until you read it closely. A majority of buyers say they’d prefer a rep-free buying experience where possible — self-service research, digital-first evaluation, minimal sales-rep contact. At the same time, 69% say they turn to a human sales rep specifically to validate AI-generated information before they trust it.

Those aren’t opposite findings. They describe buyers who want AI to do the volume work — surfacing information, answering routine questions, cutting friction out of research — while still wanting a human accountable for whether that information is actually right when it matters. Buyers don’t object to AI touching their buying process. They object to nobody being answerable for what it produces.

What Validity found among general consumers

Validity’s spring 2026 survey of roughly 500 marketers and 1,000 consumers adds a second layer, from the recipient’s inbox rather than the buying committee. A meaningful share of consumers say they’d trust a sender’s email less specifically upon learning AI wrote it — a real, if modest, trust cost tied to disclosure rather than content quality alone. A smaller but notable share have already turned off AI-related email features entirely. And a majority now report making decisions — skip, delete, open — based on an AI-generated summary of an email rather than reading the message itself.

That last point complicates the whole picture in an interesting way: the same survey population that says it trusts AI-written email less is also increasingly letting AI decide, on its behalf, whether a human-written email is even worth opening. The suspicion of AI on the sending side and the reliance on AI on the receiving side are happening in the same inboxes, often to the same people.

Finding	What it actually says
69% of B2B buyers (Gartner)	Validate AI-generated info with a human rep before trusting it — not a rejection of AI, a demand for accountability
Buyers prefer rep-free buying (Gartner)	True at the same time as the above — buyers want AI for volume, humans for verification
Consumers trust AI-written email less (Validity)	Tied to disclosure/knowledge that AI wrote it, not to a measured drop in opens or replies
Consumers act on AI email summaries (Validity)	Majority now skip/open/delete based on a summary, not the message itself

What this implies for the AI-vs-human-review question

Put together, these two surveys argue against both extremes in the “how much should I let AI generate” debate. Fully AI-generated, unreviewed outreach runs straight into the trust cost Validity measured — recipients who find out (or suspect) content is unverified AI output discount it. Fully human, unassisted outreach ignores the volume and research advantage buyers themselves say they’re fine with AI providing, per Gartner’s findings on where buyers already welcome AI in their process.

The pattern both surveys point to is closer to what buyers actually asked for: AI doing the research, drafting, and volume work, with a human checking and standing behind what actually gets sent. That’s not a hedge — it’s the literal thing 69% of buyers in the Gartner survey said they want when AI is involved in something they’re deciding whether to trust. AI SDRs vs. human judgment covers where that split concretely falls in a cold email workflow: what’s safe to automate and what isn’t.

What this looks like inside an actual sales process

Picture a rep working a list of two hundred accounts. AI drafts the first-touch email for each one, pulling from firmographic data and recent company news to make the opener specific rather than generic. Under the Gartner finding, that’s the part buyers are fine with — nobody surveyed said they mind AI doing the research and the first draft. Where the same buyers start reaching for a human is later in the conversation: once a prospect asks a pointed question about pricing, an implementation timeline, or a claim the email made, that’s the moment 69% of buyers say they want a person who can stand behind the answer, not a second AI-generated response. The dividing line isn’t “did AI touch this email” — it’s “does a human show up before the buyer has to decide whether to trust what was said.”

A few things worth clarifying

If AI drafts the email, does disclosing that hurt reply rates?+

The Validity data measures trust in general commercial email, not response rates on cold outreach specifically, so treat any direct claim about reply rates as unverified. What the survey does support is a narrower point: recipients react to feeling unverified, not to AI having been involved at all. A reviewed, accurate, relevant email doesn't carry the same trust cost as one that reads as generic or gets a fact wrong — disclosure on its own isn't really the variable buyers are reacting to.

Is a human review step actually enough to satisfy what buyers are asking for?+

Per the Gartner finding, yes, directionally — the behavior buyers describe is turning to a human to validate AI-generated information, not rejecting AI-touched content outright. A meaningful human check before send is the same shape of accountability buyers say they already seek out later in a deal. It isn't a guarantee every recipient will react well, but it's the specific gap the survey data identifies, and closing it is a defensible standard rather than a guess.

The practical takeaway

If you’re deciding how much to lean on AI generation before a human reviews a send, these two surveys give a concrete answer rather than a vague “be careful”: buyers are not rejecting AI-assisted outreach. They’re rejecting outreach nobody verified. The fix isn’t turning AI off. It’s making sure a human is still the one who can say, honestly, that they checked what’s about to go out — which is a cheaper and more durable standard to hit than trying to make AI-generated copy undetectable.

That’s the specific split Norbelys is built around rather than treated as an afterthought: AI drafts a sequence from a campaign brief, but nothing goes out until a human reviews it, matching the human-in-the-loop guardrails on the sending domain both surveys argue buyers actually want. Norbelys doesn’t send unreviewed AI copy on autopilot, and it doesn’t ask you to write every sequence from scratch by hand either — the same 69% of Gartner’s buyers who say they want a human to validate AI-generated information are describing the exact checkpoint a review step before send is meant to satisfy.