The Guardian may not be entirely blameless in this respect, but Ben Goldacre calls three recent egregious pieces of fiction in other newspapers that masquerade as reports of science findings:

• The Telegraph ran a headline "Wind farms blamed for stranding of whales", but the report that it referred to made no mention of wind farms (and the report has now been retracted).
  
• The Telegraph ran "Why stilettos are the secret to shapely legs", the Mail, "Stilettos give women shaplier legs than flats" and the Express, "Stilettos tone up your legs". The paper that was the purported source of these fables made no mention of shoes of any description, much less stilettos.
  
• The Mail ran "Swimming too often in chlorinated water 'could increase risk of developing bladder cancer', claim scientists" from a report that contained no measures of bladder cancer incidence.

Goldacre surmises that if journalists made a practice of providing links to their primary sources (as in fact do publications such as New Scientist), they would think twice about concocting such beat-ups. I'm not so sure. Newspaper editors like scare headlines and favour writers who can produce them.