Skip to main content
News Plus 5 Jul 2026  - 8 min read

HubSpot beats an immediate retreat. A week after flipping its rules and shifting beyond permission marketing to 'shared data enrichment' – with default opt-in – the white flag goes up

By Andrew Birmingham - Martech | Ecom | CX editor

Oh that policy change... Duncan Lennox, HubSpot Chief Product and Technology Officer

Would you approve of a SaaS vendor using some of your customer-derived data to improve records held by other companies - including competitors? HubSpot sure thought so. It announced new terms on 1 July that could allow enrichment data from one company to be added to HubSpot’s commercial dataset and used to enrich or supplement another customer’s records. Then it decided – arbitrarily – to opt everyone in. Quelle horreur.  The company first tried to defend itself, arguing that sensitive CRM content such as notes, deals, call recordings and custom fields sat outside the enrichment system. But that still left a veritable cornucopia of data including CRM contact records, employer information and even email deliverability signals. Sure brands could opt out, but super admins would need to navigate multiple paths to do so. By this morning, the policy was dead in the water - probably.

What you need to know

  • HubSpot has reversed controversial AI data-sharing terms just a week after announcing customers would be opted in by default to data enrichment and AI training unless super admins actively opted out.
  • Chief Product and Technology Officer Duncan Lennox admitted the rollout was a mistake, apologised for falling short on transparency and confirmed the July 1 terms changes will be completely withdrawn.
  • The backlash centred on fears customer data could help competitors, with critics arguing HubSpot had shifted from a permission-based CRM to a shared commercial data model without explicit customer consent.
  • HubSpot maintained core CRM content was never in scope, but proposed terms allowed defined business contact and company data to be added to a commercial enrichment dataset used across customers.
  • AI training and enrichment were separate opt-out settings, adding to customer confusion over what data would be shared, used to train models or contribute to Contact Discovery launching August 4.
  • Partners, consultants and customers challenged inconsistencies between HubSpot's long-standing claims that customer data is never shared and the new "give-to-get" enrichment model.
  • Competitors seized on the controversy, with some comparing HubSpot's approach to ZoomInfo's data aggregation model and accusing the company of abandoning its permission-first brand positioning.
  • The episode echoes Salesforce's earlier AI opt-out controversy, but critics argue HubSpot went further by explicitly proposing customer-contributed data be used to enrich other customers' records.
  • Australian brands may face privacy compliance questions if business contact data collected for sales or marketing is later used for AI training or enrichment without disclosures aligning with the Privacy Act and Australian Privacy Principles.
  • The rapid policy reversal highlights growing scrutiny of enterprise AI data governance, with vendors increasingly forced to balance AI ambitions against customer trust, transparency and consent.
  • HubSpot's retreat doesn't end the issue says Civic Data's Chris Brinkworth, He argues the company has paused its AI data-sharing changes, but the legal and privacy questions remain unresolved, with HubSpot signalling an opt-in version will return.
  • Australian brands may carry the risk because when CRM data collected for sales and marketing ithat s later used for AI training or data enrichment, organisations – not just vendors – could face Privacy Act, spam and AML compliance issues if disclosures, consent and governance don't match actual data flows.

Your data belongs to you. That's always been our commitment, and it won't change. We still believe there's a better way to prospect than the status quo. But we have to earn your trust as we build it. We'll take the time to do this right.

— Duncan Lennox, chief product and technology officer, HubSpot

That was fast! HubSpot tried to flip the rules on its customers, by turning permission on by default to use those customers' data for AI training and enrichment unless super admins opted out. In practical terms, that meant that a brand's data might be used to enrich a competitor's data.

The policy lasted barely a working week. This morning, HubSpot chief product and technology officer, Duncan Lennox bid a hasty retreat - for now.

According to Lennox, "I want to share something with our customers directly. Last week, we announced changes to our terms of service that we should not have rolled out the way we did. We made a mistake, and we're reverting those changes completely."

The 1 July legal update triggered a backlash from partners, consultants and industry watchers, including a claim by competitors that the marketing software company has shifted beyond a permission-based platform into a data-pooling business without requiring affected customers to make a fresh affirmative choice.

HubSpot tried to assuage fears by saying notes, deals, call recordings, custom fields and other internal CRM content are outside the enrichment system. But its updated product-specific terms said customers agree HubSpot may “add Enrichment Data to our commercial dataset” and use it to supplement other customers’ data.  And HubSpot’s own AI settings page says the AI Model Training switch is “toggled on” by default.

The company said the setting controlled whether HubSpot could use an account’s customer data to train HubSpot AI models and that, when it is turned off, the data would not be used to train those models and is not shared with other HubSpot customers.

But today on LinkedIn Lennox acknowledged the mistake and retracted. "I've written about what we were trying to do, where we went wrong, and how we'll do better from here, and you can read that below. The short version: our intent was to help customers prospect in a more trusted, effective way. But we didn't meet the transparency standard you expect from HubSpot, and I'm sorry for that," he stated. 

"Your data belongs to you. That's always been our commitment, and it won't change. We still believe there's a better way to prospect than the status quo. But we have to earn your trust as we build it. We'll take the time to do this right."

Rival tech player, Salesforce faced a similar opt-out controversy earlier this year, with its own help pages saying customers can exclude their data from Einstein global models. But HubSpot’s change went further by expressly allowing defined enrichment data to feed a commercial dataset used to supplement other customers’ records. 

According to the CRM vendor, its Contact Discovery, a feature launching 4 August, will let customers find, verify and add new contacts inside HubSpot. The company said the feature powered by a shared dataset and that “some of your business contact data helps keep that dataset current.” Opt-outs would require action by super admins.

The company's community forum revealed for CRM contact records, the data in scope could include professional details such as a person’s name, work email address, job title, employer, role, seniority, business location and profile URL. HubSpot said it may also use enrichment-related metadata, including whether a record has been enriched or whether it has been opted out. For company records, the relevant data could include the organisation’s domain, name, industry, location, employee count, size, revenue, description and related profile information.

That opt-out language has also fuelled the criticism. HubSpot’s opt-out page said its AI models use customer data “to provide better results to all users,” while also saying data is never shared between users or accounts.

On the community forum, one user said the company’s email warned that enrichment data such as business contact details, employer information and email deliverability signals “may be shared with other customers” from 4 August. The user asked whether AI model training and enrichment sharing were separate settings and whether both had to be turned off. Short answer - yes.

On LinkedIn, the occasionally aggrieved and confused comments by customers and partners were already enriching HubSpot's understanding of how the policy change was being received – and you didn't need your super-admin to tell you the answer is "poorly".

Clean up in aisle 9

Before the reversal, the firm's chief customer officer John Dick was already trying to hose down concerns, saying the CRM vendor was focused on "helping customers get data that’s current, validated, and actually useful, with better signals for fit and intent."

As to what was changing, he wrote, "HubSpot has always collected name, work email, and company domain from enriched records. Starting 4 August, we’ll also include professional details like job title, role, seniority, business location, and professional profile URL. Basically, the kind of information you’d find on a business card. These fields are also available to enrich. It works on a give-to-get model. If you use HubSpot’s enrichment, you contribute “business card-level” data to help keep the dataset accurate and up to date. In return, you get access to better, more reliable data over time. That shared model is what makes the system more valuable for everyone who participates."

He argued customers are still in control of their data. "What’s in your portal is still yours. Your CRM data, notes, deals, call recordings, custom fields, and internal content are never part of enrichment and are never shared. And if you don’t want to use or participate in enrichment, you can turn it off at any time. We believe in customer choice."

Ruffled feathers

But that's triggered dissent and scepticism within the community.

Noting that a HubSpot Knowledge based article claims “Data is never shared between HubSpot users or accounts,” Thomas Ekers, a London based solutions architect, wrote, "That’s confusing. If data is used to improve results for all users, the distinction between “used” and “shared” needs to be much clearer. He then linked to a HubSpot security article, which also stated  “We never share your data across customers.”

Per Ekers: "Given the give-to-get model, that no longer seems accurate. Even here, your explanation says CRM data is never shared, but that doesn’t make sense if it’s used to improve results for other users. Moving from 'we never do this' to 'we’ll do this for one service unless you opt out' is bound to ruffle feathers, especially for people who bought/configured HubSpot around that stance."

Meanwhile, competitors made hay. David Elkington, the founder and CEO of Atonom, which integrates into HubSpot, but is also a direct competitor to HubSpot Breeze AI - described the new approach as a "desperation move" and wrote, "If your account already uses enrichment, you are enrolled by default. HubSpot calls it admin-controlled. But it isn't one switch, ... admins on Reddit have mapped at least 3 separate toggles: (1) enrichment sharing, (2) tracking-code intent, and (3) AI training on sent emails. All tied to Product Terms and a DPA that changed 1 July."

Elkington argued the move was a response to HubSpot's tanking share price - part of the so called SaaS-apocalypse - "When markets stop paying you for software growth, you find something else to sell. And the most valuable asset inside HubSpot is not Breeze. It is the contact and engagement data sitting in a couple of hundred thousand customer accounts."

He described this as "ZoomInfo's playbook, ... pool the customers' data and sell it back as intelligence. Except ZoomInfo customers knew exactly that trade going in. HubSpot built its entire brand on the opposite pitch, they literally wrote the book on inbound and permission."

Independent analysis using the Value Gravity framework designed by Arjen Segers, a Dutch independent advisor, noted that the legal groundwork was already in force, but the operational change was scheduled for 4 August. It confirmed that not all CRM data was affected, but that it could include business contact data, company data, tracking-code data and email engagement signals.

The individuals whose details fill these CRMs were mostly never told; and we don't feel that any of the regulators will agree that an 'admin toggle' is considered consent.  The most durable risk is a businesses own words and a policy saying 'we never share your data' while signals feed a shared dataset. That is a gap that that business/HubSpot user owns, whatever HubSpot does next. 

— Chris Brinkworth, Director, Strategic Advisory, Civic Data

Australian privacy laws

For Australian customers, HubSpot’s changes raise questions about whether brands have properly disclosed how customer and prospect data may be used once it enters the platform. Business contact data can still be personal information under Australian law if an individual is reasonably identifiable. That means names, work email addresses, job titles, company affiliations, email engagement signals and website behaviour may fall inside the Privacy Act, depending on the context. The Australian Privacy Principles require organisations to be open about how they handle personal information and generally restrict use or disclosure to the purpose for which it was collected unless an exception applies.

That could create exposure for Australian brands if data collected for sales, service or marketing is later used to train vendor AI models, improve enrichment products or supplement another customer’s records. The practical question for CMOs is whether their privacy policies, collection notices, consent records and vendor governance processes match the data flows now described in HubSpot’s terms.

According to Chris Brinkworth, Director, Strategic Advisory, Civic Data, which advises Australian companies on the technology and regulation behind their customer data systems, "Our read: the walk back closes the 'terms', but not the questions that people should be asking. It was customer outrage, not any regulator, that forced this reversal inside a week; nothing was legally tested, and HubSpot still says "the model will return as opt in".    So our view is that the questions are still live, and they apply to any vendor running enrichment and not just HubSpot. 

He said privacy law generally stops information collected for one purpose being used for another unless an exception applies; "The individuals whose details fill these CRMs were mostly never told; and we don't feel that any of the regulators will agree that an 'admin toggle' is considered consent.  The most durable risk is a businesses own words and a policy saying 'we never share your data' while signals feed a shared dataset. That is a gap that that business/HubSpot user owns, whatever HubSpot does next."

He said, "Those businesses with the most at stake are actually the ones assuming this is someone else's problem: much of HubSpot's base is small to mid-sized and relies on the Privacy Act's small business exemption.   But the regulators likely to move first have no size test.  Spam law applies from the first marketing email, and the sender must prove consent, no matter who supplied or 'verified' the contact, and since 1 July, the second phase of Australia's anti-money laundering laws has pulled thousands of smaller firms, real estate agents, lawyers and accountants among them, into obligations that also switch off that exemption under the Privacy Act."

In other words, "We're too small for privacy law' is now the least safe assumption in Australian marketing.

Per Brinkworth, "So while HubSpot says it is walking everything back, our read is that three things remain true. Going back to the old terms is not going back to zero: HubSpot itself says it has always collected names, work emails and company domains from enriched records, and that baseline sharing carries on as before. The company has also been clear the model will return, rebuilt so customers actively choose to take part. And because customers, not regulators, forced the retreat, none of the legal questions were ever tested.  This backdown bought time, not answers.   Brands that use that time to check their privacy notices, consent records and vendor settings will be ready when this idea comes back. And it will."

What do you think?

Search Mi3 Articles