There are a number of key differences, however. Unlike SOPA, PIPA lacks a provision that requires search engines (like Google and Bing) to remove a “foreign infringing site” from their indexes. This provision in SOPA is one of the most highly criticized.

PIPA also contains provisions that require greater court intervention to go after an accused website than SOPA does. But it does not contain any provisions that would penalize copyright holders for misrepresenting the alleged infringing activities of an accused website — a potentially disastrous omission for innocent sites put through the PIPA ringer. SOPA does contain a provision that penalizes those copyright holders who “knowingly materially misrepresent” the alleged infringement of a website by making them “liable for damages, including costs and attorneys’ fees, incurred by the person injured by such misrepresentation as a result of the misrepresentation.”
[..]
Also included in both bills is an anti-circumvention provision, which would make it illegal to inform users how to access blocked sites. According to First Amendment expert Marvin Ammori, a Legal Fellow with the New America Foundation Open Technology Initiative and an Affiliate Scholar at Stanford Law School’s Center for Internet & Society, this provision could be interpreted to require any website that features user-generated content (think Facebook, Reddit, or YouTube) to make sure that no circumvention information has been posted to their site, or risk legal action.

Another provision highly cited by critics is the “vigilante” part of SOPA/PIPA, which allows ISPs to voluntarily block access to certain foreign websites, “in good faith,” if they have “credible evidence” that these sites are devoted to illegally distributing copyrighted material. Both SOPA and PIPA give immunity to ISPs who take voluntary action against websites that are believed — but not proven — to be dedicated to the illegal distribution of intellectual property. Because of this, critics say the potential to abuse this power is unconscionable.

An analysis (PDF) of Protect IP prepared by five Internet researchers this spring lists potential security problems. Among them: it's "incompatible" with DNSSEC, innocent Web sites will be swept in as "collateral damage," and the blacklist can be bypassed by using the numeric Internet address of a Web site. The [IP] address for CNET.com, for instance, is currently 64.30.224.118.
[..]
A little-noticed portion of the proposed law, which CNET highlighted in an article, goes further than Protect IP and could require Internet providers to monitor customers' traffic and block Web sites suspected of copyright infringement.

"It would cover IP blocking," says Markham Erickson, head of NetCoalition, whose members include Amazon.com, Google, eBay, and Yahoo. "I think it contemplates deep packet inspection" as well, he said.
[..]
Deep packet inspection, meaning forcing an Internet provider to intercept and analyze customers' Web traffic, is the only way to block access to specific URLs.