Improved AI-Powered Gibberish Detection for Form Submissions

Chris drove the long conversation here, and Trisha supplied the operator's reason to care. She picked this one first. At every company she has worked for she has seen waves of spam arrive through specific campaigns, she said, and that goes for anyone running Google Ads. It does not hit consistently across the board: it shows up in one area and then suddenly junk is coming through. It is entered so that it passes the normal sniff test. A human can look at it and say that is not real, but to the way HubSpot reads the data it looks good enough. Her stance on the update: we will see how good it is.

Chris read the update as written, that AI-powered gibberish detection now covers all text fields on a form submission, not just first name, last name and message, and went to the bigger picture. As he sees it, a spam bot is basically an agent on the other side now, so we are expecting AI technology to combat it. At some point, he said, forms may not be the best thing to count on for conversion information. That takes him to signals and custom signals: understanding where somebody was and reaching them a different way before they get to the form, which he thinks is a real opportunity for companies that have been relying on lead magnets to do the work. He hoped HubSpot could use this to produce insights, such as “maybe you should shut this form down” because it is a hotbed for spam, so get more specific about which page it is on or start protecting it.

Trisha pushed back on exactly that. Asking companies to drop the contact form is a revolutionary idea, she said. Everybody expects a way to contact a company, removing it weeds out the bad with the good, and there is no vision yet for what the replacement experience looks like. Chris said he is not saying get rid of it, but that it could look different: not just a hands-off form that asks twenty questions so it knows which rep to route you to by zip code or region, but meeting booking links and smart content that adjusts by IP, for example sending someone in Texas to Chris and someone elsewhere to a round-table booking link. Trisha had seen booking links get spammed too. Chris also pasted a screenshot into Breeze Assistant and asked what a team could learn from spam submissions; Trisha had not known you could paste images in there now. His closing read: tightening prevention at the form level is where everyone ends up, and it starts to make things hard for the real people who want to submit, so usually you make the business case to just deal with the spam submissions, and that is what this update helps with.

What to do: this touches Marketing Hub and Content Hub Professional and Enterprise portals, in public beta, and sensitive data fields are excluded. If you run paid traffic to forms, watch which forms the spam waves hit and what the detection actually catches before you rely on it. Whether your form should keep its current shape is Chris's question to sit with, alongside Trisha's caution that no replacement is ready yet. Trisha's own early note: she may need a new naming convention for her test submissions.