Article 13 of the Copyright Directive requires online services to crowdsource a database of "copyrighted works" (anyone can add anything to these databases, with no penalties for falsely claiming copyright over public domain works, or works that don't belong to you).
If a user tries to post something that appears to match an item in the blacklist database, the service has to censor that user's post.
The problem is that the online platforms don't have "European services" and "non-European services," they just have services, and users from the EU and outside of the EU freely mix there, posting material and commenting on it.
So if you're Twitter and one of your US users posts something that has to be censored in Europe, you can either censor that just for EU users, or for the whole world. If you do decide to just censor the post in the EU, then you have users in (say) Canada replying to that US post, and now you're showing your EU users the replies to the post, but not the post itself. Indeed, you might get Europeans posting things that can't be displayed in Europe, but which the rest of the world is allowed to see -- do you censor those posts in the EU (including from the people who made them) or in the whole world?
To make this all the scarier, the EU rules provide for massive damages if a platform lets a copyrighted work slip through the cracks -- so in the event that a platform isn't sure if a work is going to be shown to a European user, they'd be nuts to take a chance on it.
The result will either be a balkanized internet where services host special versions of their conversations for European consumption (much like the censored search engine Google wants to sell to China) or an internet where we race to the bottom, as defined by European copyright extremists.
Now, let's take a hypothetical Twitter discussion between three users: Alice (an American), Bob (a Bulgarian) and Carol (a Canadian).
Alice posts a picture of a political march: thousands of protesters and counterprotesters, waving signs. As is common around the world, these signs include copyrighted images, whose use is permitted under US "fair use" rules that permit parody. Because Twitter enables users to communicate significant amounts of user-generated content, they’ll fall within the ambit of Article 13.
Bob lives in Bulgaria, an EU member-state whose copyright law does not permit parody. He might want to reply to Alice with a quote from the Bulgarian dissident Georgi Markov, whose works were translated into English in the late 1970s and are still in copyright.
Carol, a Canadian who met Bob and Alice through their shared love of Doctor Who, decides to post a witty meme from "The Mark of the Rani," a 1985 episode in which Colin Baker travels back to witness the Luddite protests of the 19th Century.
Alice, Bob and Carol are all expressing themselves through use of copyrighted cultural works, in ways that might not be lawful in the EU’s most speech-restrictive copyright jurisdictions. But because (under today's system) the platform typically is only required to to respond to copyright complaints when a rightsholder objects to the use, everyone can see everyone else's posts and carry on a discussion using tools and modes that have become the norm in all our modern, digital discourse.
But once Article 13 is in effect, Twitter faces an impossible conundrum. The Article 13 filter will be tripped by Alice's lulzy protest signs, by Bob's political quotes, and by Carol's Doctor Who meme, but suppose that Twitter is only required to block Bob from seeing these infringing materials.
EU Internet Censorship Will Censor the Whole World's Internet [Cory Doctorow/EFF Deeplinks]