Texas Attorney General Ken Paxton has put data-hungry businesses on notice: Give consumers a fair shake or pay the price   What is your customers’ data worth to them? It’s the key question behind Paxton’s new lawsuit against Allstate and its subsidiary Arity. The suit claims that Arity monitored driver behavior by tracking users' phones, then sold that data on to insurers, who used it to jack up motorists’ insurance costs or deny them coverage. Allstate, naturally, contests the claims. The underlying issue, though, is that drivers had no idea their data was being mobilized against them, and appear to have received little of value in exchange.   Allstate isn’t alone in avidly collecting data, of course. In the AI era, businesses of all kinds are gathering all the data they can, and using it to build the AI models they’re counting on to deliver competitive advantage. The problem is that leaders focus so much on the value data has for their business that they overlook the fact that it also has value to their customers. All too often, that leads them to see data collection as a one-sided, extractive process, rather than an equitable exchange that benefits both customers and businesses.   To put it another way: data collection doesn’t just generate value for the company that gathers it. It also imposes a cost on the data subject. (In the case of Allstate, it was a literal cost in the form of higher insurance premiums.) Unless consumers feels they’re getting a fair shake and receiving something of equivalent value in exchange for their data, they’re quite rightly going to resent being taken advantage of. In the long run, that will cause all kinds of problems — including reputational harm and heightened regulatory attention — as companies seek to develop effective and sustainable AI strategies. It might sound strange to talk about data collection as imposing a cost on consumers. When I share my location or let someone track my browsing, have I really given up anything of value? Some experts argue otherwise, but what really matters is that consumers think their data is valuable. In fact, three-quarters of consumers say they highly value their data privacy — and while four-fifths acknowledge that there can be a potential upside to sharing data with businesses, that doesn’t mean they’re willing to share any data, with any business, for just any reason. Instead, research shows that consumers carefully weigh the costs and benefits when granting access to their data. Most consumers will happily hand over their purchase history in exchange for more personalized shopping experiences, for instance. Far fewer, though, will grant access to more sensitive information, like the contents of private emails, in exchange for discounts or perks.   In practical terms, that means that consumers view data-sharing as a direct cost that needs to be offset by the benefits provided in exchange. And they resent it when companies try to simply grab their data without offering something worthwhile in return.  This doesn’t mean, as some have suggested, that consumers should be paid for their data. In fact, research shows that people actually trust companies less when they try to buy access to their data — the attempt always feels a bit too grubby and transactional. But it does mean businesses should view data collection in terms of the return for their customers, and make sure they’re delivering real value in exchange for the data they demand.  Some companies are already doing this. Apple, for instance, was a relative latecomer to the AI party, but the Cupertino company is charting a course that hews pretty close to where consumers currently stand — and communicating clearly with customers about how its on-device and “private cloud” tools enable AI while keeping their data safe. By keeping the costs low relative to the value delivered, Apple is able to deliver significant return on investment for its users.  Other companies, of course, take a different approach. X.com recently revamped its privacy policies, and is now freely shoveling user data into its AI models. That imposes new costs on users — and if customers don’t get significant new value in exchange, the “Twitter exodus” might continue to snowball.   The point here isn’t that businesses shouldn’t collect data. It’s that data collection is a price signal — a cost imposed on customers that needs to be offset by the value provided. If you’re only collecting trivial kinds and amounts of data, you can get away with offering trivial returns. If you’re collecting data that consumers perceive as more valuable — like medical records, say, or user-created content — then you need to mitigate that cost with robust data protection, and ensure you’re delivering truly compelling value to the user.   A couple of years ago, the World Economic Forum proposed trading data much as we trade stocks, complete with “data exchanges” to price and trade data products. But individuals also need a clear acknowledgement of the value exchange they’re participating in whenever they grant access to their data. Putting consumer return on investment at the center of data collection would emphasize the often-overlooked truth that data subjects are free agents, with the right to demand real value in exchange for their data.   Doing so would empower consumers, while also helping businesses to thread the needle between rapid AI innovation and responsible (and legally compliant) growth. By emphasizing return on investment, companies naturally align themselves with core ethical data principles.   Data minimization, for instance, is really a way of ensuring that costs to data subjects are proportional to value provided. Transparency and consent work the same way: companies must provide enough information and choice to minimize the cost of data collection, without overwhelming users with legalistic privacy language and intrusive consent requests.   In the long run, companies that fail to strike this balance will pay the price. In some cases, regulators — in both red and blue states, and federally — will step in, and demand that companies delete improperly collected data or disgorge improperly trained AI models. In others, users themselves will flex their muscles and migrate to AI providers that minimize the perceived cost of data collection while maximizing the benefits provided in exchange.   As the AI ecosystem matures, the winners won’t be the companies that relentlessly harvest data from their users. Instead, they will be the companies that respect their users, and ensure the data they collect is truly proportional to the value they deliver.  Jonathan Joseph is a board member of The Ethical Tech Project.