'Surveillance pricing': Why you might be paying more than your neighbour (www.aljazeera.com)

Retailers and travel companies are increasingly using AI-driven “surveillance pricing” — mining web pixels, login and purchase histories, device signals (even battery level), mouse movements and video engagement — to estimate each customer’s willingness to pay (WTP) and adjust prices in real time. The trend hit headlines after Delta Air Lines said in July that roughly 3% of its domestic fares are now set using AI, with a goal of 20% by year-end, prompting a congressional inquiry; regulators such as the US FTC have documented how firms segment customers and test price points to maximise revenue. High-profile investigations and experiments (including reports linking low phone battery to higher ride prices) show this isn’t hypothetical: intermediaries can aggregate and sell WTP signals, and algorithms can move from broad store-level strategies to per-user price recommendations. For the AI/ML community this raises technical, ethical and regulatory challenges. Models used to infer price sensitivity rely on large, often opaque feature sets and continual online experimentation, amplifying risks of unfair discrimination, privacy invasion and non-transparent decisioning. Policymakers are responding — New York banned undisclosed personalised algorithmic pricing, Ohio and other states seek disclosure requirements, and the UK’s DMCCA enables heavy fines for biased digital pricing — pushing practitioners to prioritise model explainability, auditing, data minimisation and compliance when designing pricing systems.