HubSpot reversed course this summer after customers pushed back on a data-sharing requirement connected to its AI-powered features. The specifics matter less than the underlying pattern, which is going to keep repeating across the martech landscape: as CRM platforms evolve from customer databases into AI systems that identify buying intent and take action, they need more of your data to do it well. Customers are increasingly asking who captures the value when that data gets shared.
The pattern behind the headline
Every AI feature bolted onto a CRM, an email platform, or an ad tech tool runs on data, and increasingly on data that used to sit quietly inside your own systems. Vendors have real incentive to pool customer data across their base to train and improve the AI features they are selling you. Customers have real incentive to keep control over data that represents their own competitive position. Those incentives are not always aligned, and platforms are still working out where the line sits.
What this means if you run marketing operations
Read the AI feature terms separately from the platform terms. Many vendors are now bundling AI-feature data usage into updates that ship separately from a core terms-of-service revision. Do not assume your last review of the platform’s data policy still covers what its AI features do with your data today.
Ask what “improve the product” means in practice. That phrase in a data policy can mean anything from anonymous aggregate benchmarking to training a model that indirectly benefits your competitors. Vendors that can answer specifically are worth trusting more than vendors that answer in generalities.
Watch for reversals as a trust signal, not a red flag. A vendor that ships a change, listens to backlash, and reverses it quickly is behaving better than one that quietly keeps the policy and hopes attention moves on. HubSpot’s reversal here is a point in its favor, not against it.
Why this is structurally different from past data-policy fights
Data-sharing controversies aren’t new to SaaS — platforms have been quietly loosening data-usage terms for years. What’s different this time is the mechanism: AI features generally need broader, more granular access to perform well than a traditional feature does. A reporting dashboard needs your metrics. An AI agent that’s supposed to identify buying intent and act on it needs your full CRM history, your email engagement patterns, and often cross-customer benchmarking data to calibrate against. That’s a categorically larger data footprint, and it’s why AI-feature rollouts keep triggering these fights even at vendors with a previously clean data-policy track record. The AI feature isn’t doing anything sneaky — it genuinely performs better with more access — which makes “how much access is too much” a much harder line to draw than it used to be.
A practical framework for evaluating any vendor’s AI data terms
When a vendor in your stack ships a new AI feature, three questions are worth asking before you opt in, not after:
Is the data used to improve my instance, or the vendor’s model broadly? The first is a reasonable trade — you get a better product. The second means your data is contributing to a system your competitors may also benefit from, even if it’s never directly exposed.
Can I opt out of the AI feature without losing the underlying platform functionality? If the AI capability is bundled so tightly that declining it degrades the core product, that’s worth flagging to your vendor management or legal team rather than accepting by default.
Does the vendor have a specific, technical answer, or a marketing answer? “We take your privacy seriously” is a marketing answer. “Your data is used only to fine-tune your instance’s recommendations and is not included in any base model training” is a technical answer you can actually hold them to.
The bigger takeaway
This will not be the last time a martech platform’s AI ambitions collide with customer data expectations. The marketing leaders who navigate it well are the ones treating data-sharing terms as a recurring line item to review, not a box checked once at signup.