The scientist reports on his work. There's a natural tendency to point up the promising aspects, and what it might lead to.
The reporter writes a story about the story. With rare exceptions, the reporter knows little about the topic (or any topic, for that matter), but does know how to quickly write an interesting narrative, and will seize on those aspects that make for a good story.
The editor decides on a headline. The editor has never met the scientist, and probably doesn't know statistics from library paste. The editor looks for something eye-catching in the story, even if meanings have to be contorted a bit and possibilities magnified still more.
Think of it as a game of telephone in which each player has to restate the message in a more dramatic way.