Netflix tests OpenAI-powered search engine for movies and shows

We have all spent forty-five minutes scrolling through streaming rows only to end up watching absolutely nothing and going to sleep.

Netflix wants to fix that. As first reported by Bloomberg, the streaming giant is testing a new AI-powered search feature to recommend movies and television shows, and it is tapping OpenAI tech to make it happen.

Searching by vibe instead of actors

Traditional streaming search has basically been stuck in the past: you type an actor, a director, or a broad genre tag like comedy or sci-fi, and you hope for the best.

Netflix confirmed that this new test allows subscribers to search using far more specific terms, including their actual mood. Instead of just picking a category, users can ask for something very nuanced that goes way beyond standard credits.

Think of it like having ChatGPT built directly into your TV app. You can ask for a classic nineties action thriller, under two hours long, with minimal violence, and receive a curated list drawn entirely from what Netflix actually has in stock.

A Netflix spokesperson confirmed the details to The Verge, calling it an opt-in beta that is currently in a learn and listen phase. For now, testing is limited to iOS users in Australia and New Zealand, with plans to roll out to US iPhone users over the coming weeks and months. There are no current plans for platforms outside iOS.

Neither Netflix nor OpenAI made a grand public announcement about the partnership, likely reflecting ongoing industry sensitivity around artificial intelligence in creative entertainment spaces.

Why it matters

Search on connected devices has barely changed since the DVD era, and turning conversational AI loose on a massive media library is an obvious upgrade.

For viewers, the real win is ending the endless decision paralysis. People already ask chatbots what to watch, but half the time those movies are scattered across different platforms you do not own. Bringing OpenAI's natural language comprehension directly inside Netflix cuts out that friction completely.

That said, conversational search carries risks. If the model hallucinates tags or misunderstands subjective emotional prompts, users will bounce back to the traditional trending row in seconds.

What is most fascinating here is Netflix's strategic choice. Despite spending two decades refining its internal machine learning and recommendation pipelines, it recognized that general-purpose language models are simply better at parsing human conversation than legacy search systems. If this lands well, every rival streaming app will have no choice but to license an LLM for their own search bar.