AI conversations are a secret new treasure trove for marketers (www.theregister.com)

Startup Profound markets a service that lets companies see how they appear in AI-driven customer interactions, boasting access to "150+ million real user conversations." But privacy experts tell a different story: Lee S Dryburgh and others allege Profound is buying clickstream data harvested by browser extensions and mobile wrappers that capture users' chatbot prompts and responses — often without clear, specific consent. Profound denies using "stolen" data and says its datasets are opt‑in and GDPR/CCPA‑compliant; the company has not clarified its data suppliers. Investigations point to potential intermediaries such as Datos/Semrush and to common extension permissions (activeTab, scripting) that can read page content and inject code, making sensitive queries about health, finance, and identity easy to collect and resell after "anonymization." The story matters because search‑style brand visibility is shifting toward AI Optimization (AIO), creating strong commercial demand for conversational data — but also major legal and technical risks. GDPR, PECR and CCPA set high bars for informed, specific consent and limit traffic‑data processing; vague extension permissions and broad opt‑ins may not satisfy them. Beyond legality, provenance and consent affect data quality and re‑identification risk, undermining the reliability of insights companies buy. Regulators and platforms may soon clamp down on clickstream collection, and brands should treat such third‑party conversational datasets with caution.