Please Stop Making Me Opt Out of AI
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Meta rolled out an AI app feature in early July that let users generate images tagging public Instagram accounts, opting users in by default; after creators posted opt-out videos—one by Sam Sooin Yang hit 3 million views—Meta reversed course in three days, calling it a feature that "missed the mark."
- Electronic Frontier Foundation's Thorin Klosowski called the rapid user-led reversal of a generative AI default "a clear and immediate pushback" and noted three days from launch to rollback may be a record for a generative-AI feature.
- Boston University law professor Woodrow Hartzog cited GDPR Article 25 as a model requiring the "more privacy protective option" to be pre-selected by default, arguing that since users tend to stick with whatever default is presented, the default choice effectively determines who gets enrolled.
- Consumer Federation of America's Ben Winters described Meta as "stewards of the opt-out status quo," noting the pattern extends to Google (which added a Gemini bar inside Google Docs), Dropbox, and LinkedIn.
- Meta also auto-enrolls users in Facebook's "Enhanced Browsing," which tracks websites visited in-app on mobile, illustrating how deeply the default-enrollment model runs across the company's product line.
- Winters argued the situation demands federal regulation, saying "public sentiment is putting us closer to federal regulation than we were a decade ago," even as prior national-level privacy bills have failed.
Why it matters: Because users overwhelmingly stick with whatever default is presented, the choice to opt users into AI features like Instagram image-tagging or Google Docs' Gemini bar isn't neutral—it shifts millions of people onto generative AI tools they never requested. The three-day Meta reversal shows public pressure can move quickly, but the broader opt-out norm across Meta, Google, Dropbox, and LinkedIn persists without federal intervention to flip defaults toward privacy-protective choices.
