We already know how the advertising model corrupted social media. Now it's heading for AI agents — and the mechanics make it potentially worse.
When an agent books your flight, recommends a tool, or summarizes your options, how will you know if that answer was shaped by a sponsored result? Unlike a clearly labeled Google ad, the influence could be baked directly into the model's training data or RAG pipeline — invisible to the user.
I wrote about where this is already happening (Perplexity's sponsored answers), how affiliate-style incentives could quietly distort agent recommendations, and what "hyper-personalized" ad targeting looks like when the agent knows everything about you.
👉 Read the full article on HackerNoon
Curious if others are thinking about this — especially those building agents that make recommendations or take actions on behalf of users.
Top comments (0)