What would you have done? — HubSpot tried to feed its AI with its customers’ data, and the revolt lasted four days

On July 1, 2026, HubSpot changed its terms of service: its customers’ enrichment data (business contacts, employer information, email deliverability signals) would be pooled to power a new AI lead-finding tool, and could be shared with other customers. The feature was on by default, leaving each customer to opt out. Three options were on the table: (A) turn it on for everyone and put opting out on the customer (opt-out); (B) turn it on only for those who explicitly agree (opt-in); (C) drop the idea of using customer data to train the AI.

HubSpot had chosen (A). Within four days, sales leaders threatened to switch providers, insisting the data was theirs and not the vendor’s. On July 5, co-founder Dharmesh Shah reversed course: “We made a mistake and are reversing that decision,” with any future use of customer data moving to opt-in. The lesson runs well beyond software: when an AI feature draws on data your customers or your teams consider their own, the default setting is not a technical detail but a trust decision, because opt-out treats consent as already given while opt-in asks for it. Before switching on such a feature, set it to opt-in, off by default, and say plainly what data is used and why.

Source: The Information / The Next Web, “HubSpot reverses customer-data AI feature,” July 2026; Salesforce Ben, “HubSpot backtracks after AI data-sharing controversy,” July 2026.