This bill pretty much demands that YouTube Content ID like systems are introduced across the web. With the wording, it's better for the site to delete more content on false positives than risk false negatives. All in the name of mighty copyright. Side effects of these false/negative averse copyright takedown systems can be easily seen on YouTube: https://news.ycombinator.com/item?id=17338700
What makes you say that? Last time I read parts of it I got the impression that if you're trying to get it right you're not liable.