From Clicks to Consensus: Collective Consent Assemblies for Data Governance
This paper proposes "collective consent assemblies" as a deliberative alternative to the flawed individual notice-and-consent model, arguing that this framework better addresses the impracticality of individual consent and the communal nature of data harms, particularly in scenarios like GenAI training.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
The Big Problem: The "Cookie Monster" is Exhausting You
Imagine you walk into a giant supermarket (the internet). Every time you pick up an item, a robot stops you and asks, "Do you want us to track your footsteps, record your voice, and sell your shopping list to strangers?" You have to say "Yes" or "No" to thousands of items every day.
This is how Notice and Consent works today. You get a pop-up (the "Notice") and you click a button (the "Consent").
The paper argues this system is broken for three reasons:
- It's impossible to understand: The rules are written in tiny, confusing legal language. It's like asking a toddler to read a physics textbook before letting them play.
- It's too much work: You get "consent fatigue." You just click "Accept All" because you're tired, even though you don't really want to.
- It misses the big picture: Sometimes, data isn't just about you. It's about a group. For example, if a camera films a street, it captures you, your neighbor, and a stranger walking by. You can't ask the stranger for permission. Also, companies often take old data to train new AI (like Generative AI) years later, but you couldn't have predicted that when you clicked "Yes" ten years ago.
The Proposed Solution: The "Digital Town Hall"
Instead of asking every single person individually, the authors propose a Collective Consent Assembly.
Think of this like a jury or a town hall meeting.
- The Setup: Instead of 10 million people clicking buttons, we randomly select a small, representative group of people (say, 100 to 200 people). This group looks like the rest of the population (different ages, backgrounds, tech skills).
- The Process: This group is invited to a "Digital Town Hall." They aren't just asked to click a button; they are educated. Experts explain how the data will be used, what the risks are, and what the benefits are.
- The Decision: The group talks, debates, and negotiates. They decide: "Do we allow this company to use our data for this specific purpose? If yes, under what rules?"
- The Result: Their decision becomes the rule for everyone. If the assembly says "Yes, but only for improving website speed, not for ads," then that becomes the default setting for all users.
Two Real-World Examples (Vignettes)
The paper uses two stories to show how this would work:
1. Fixing the "UX Improvement" Pop-ups
- The Problem: Websites constantly ask for permission to track your clicks to "improve user experience." Most people just click "Accept" without reading, or "Reject All" and break the site.
- The Assembly Solution: A town hall is held. The website companies explain, "We need click data to make buttons bigger and faster." The citizens ask, "Are you also using this to show me ads?" The group debates and decides: "Okay, we consent to tracking clicks only to make the site faster, but we strictly forbid using that data for advertising."
- The Outcome: The website updates its settings automatically to match this rule. No more annoying pop-ups for anyone.
2. Training AI with Old Photos
- The Problem: A new AI company wants to use millions of photos uploaded to social media years ago to train a new image generator. The original users never agreed to this because the technology didn't exist yet. Also, some photos have people in them who never made an account.
- The Assembly Solution: A town hall is held, including people who did upload photos and people who didn't (to represent the bystanders). They learn how the AI works. They debate the ethics.
- The Outcome: The group decides, "No. The original consent didn't cover this, and it's unfair to use photos of people who didn't sign up. We reject this." The AI company cannot use the data.
How Do We Get There? (The "Backcasting" Plan)
The paper admits this is a big change. To make it happen, they suggest four shifts:
- For Governments: They need to update laws. Instead of forcing companies to show pop-ups, laws should allow these "Town Halls" to set the rules.
- For Companies: It might actually be easier for them. Instead of fighting with millions of angry users and lawyers, they just follow the rules set by the Assembly. It's a "safe harbor" for compliance.
- For Society: We need to stop thinking of privacy as a solo act (like wearing a seatbelt) and start thinking of it as a community act (like voting or unionizing).
- For Users: We need to trust that our "digital neighbors" in the Assembly will make good choices for us, just like we trust our elected officials.
The Catch (The Downsides)
The paper is honest about the risks:
- Loss of Individual Control: If the group says "Yes," but you personally want "No," you might feel your voice is lost. The paper suggests you should still have a way to opt-out if you really disagree.
- Manipulation: Companies might try to trick the Assembly members, just like they try to trick us with pop-ups. The paper says the Assembly must be run by neutral experts, not the companies themselves.
- Time: It takes time to hold these meetings and educate people, unlike a quick click.
Summary
The paper argues that the current "click to agree" system is failing us. It's too confusing, too slow, and doesn't work for complex situations like AI or public surveillance.
Instead, they propose a Digital Town Hall. By letting a representative group of people learn, debate, and decide on data rules, we can create fairer, clearer, and more meaningful consent for everyone. It's moving from a world of Clicks (individual, mindless) to Consensus (collective, thoughtful).
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.