> In March, the EU's civil service published details of the current voluntary arrangement, which noted that "terrorist content is most harmful in the first hours of its appearance online".
This makes me think it may be timed from the upload, not the report.
I can't even begin to imagine what repercussions that will have.
> In addition to referrals, internet companies should implement proactive measures, including automated detection, to effectively and swiftly remove or disable terrorist content and stop it from reappearing once it has been removed.
To me, that sounds a lot like they're willing to mandate AI detection in order to "solve" the "team is asleep" problem. They suggest this might be yet another rent-seeking opportunity for big internet platforms:
> To assist smaller platforms, companies should share and optimise appropriate technological tools and put in place working arrangements for better cooperation with the relevant authorities, including Europol.
No doubt, some "helpful" authorities will provide an "appropriate technological tool" to help. No need for the government to break Perfect Forward Secrecy if every site is giving it clear-text access to all uploaded content.
But that problem also basically exists for smaller sites under the "1 hour after reported" rule, too. IMO, it would basically force them to hide anything that was reported immediately, and then restore it after it was moderated.
That will make it horrifyingly easy for internet trolls to mess with things.