The amount of terrorist content online is decreasing but EU governments are still calling for upload filters, reveal internal documents released by the EU Commission this week. 
Hardly any Member State reported an increase in terrorist content online when asked by the Commission in 2018. Many attributed the decline in numbers to the increased industry efforts to remove such material as well as the demise of the “Islamic State” organization in Syria. Europol meanwhile confirms that “[i]n 2018 the so called Islamic State terrorist organisation (IS) experienced a retraction in the organisation’s media production and outreach capabilities.” 
There is ongoing controversy over the added value of proposed legislation on “preventing terrorist content online” which is currently being negotiated. While a surprising number of Member States was sceptical in 2018 regarding the idea of legislating on terrorist content online, some of them seem to have changed their view by later supporting the Commission’s legislative proposal (and the favourable General Approach) in Council.
The released documents reveal broad support among Member States for the introduction of anti-terror upload filters even though they are prone to overblocking legitimate content. In industry statements also released this week,  Google cautions against mandatory upload filters, citing an erroneous removal of a parliamentary debate on torture on Youtube as well as of activist videos on war atrocities in Syria. The Internet Archive’s statement concurs: „There is no shortage of examples of false positives identified and taken down by automated filters developed and run by the largest, wealthiest online platforms.“ There is a mention of the fact that the “hash database” used by industry for filtering can easily be circumvented by modifying images and videos. Snap sums it up as follows: “It is important that the identification of illegal content is not privatised by stealth. The detection, remediation and prevention of criminal activity is one of the key functions of the state.”
The Internet Archive also warns about excessive removal requests. Even trusted flaggers, including government authorities, have occasionally flagged parodies of extremist content or historical material archived for scientific purposes, they warn. In 2019 they went public on a particularly excessive removal request by French authorities: https://blog.archive.org/2019/04/10/official-eu-agencies-falsely-report-more-than-550-archive-org-urls-as-terrorist-content/ In the ongoing legislative negotiations there is controversy on whether any national authority (including for example Hungary’s) should be given the power to have allegedly terrorist content removed wherever it is hosted – despite varying standards among EU governments in respecting the freedom of expression and of the media.
Some governments have objected to the full release of their statements (France, UK, Slovenia, Malta). For example, information given by Slovenia on erroneous removals and information on company removal practices have been removed. The Commission argues that transparency on removals would help terrorists find suitable hosters. 
In the meantime the European Parliament has resumed its meetings on the controversial “Regulation on preventing terrorist content online” proposal. 
 Government and industry statements on terrorist content online
released this week: https://www.patrick-breyer.de/?p=590541&lang=en#documents
 Europol report on decline of IS terrorist content online: https://www.europol.europa.eu/publications-documents/eu-iru-transparency-report-2018
 Blogpost on the trilogue negotiations: https://www.patrick-breyer.de/?p=590541&lang=en