Twitter whacks 270,000 terror accounts, majority flagged by internal tools
But warns political meddling risks damaging free speech
Twitter removed more than 270,000 accounts for terrorism-related violations in the last six months of 2017 – and most of these were detected by internal tools.
According to the anger platform's latest transparency report, which covers 1 July to 31 December 2017, this takes the total removed to date to 1.2 million.
In the most recent batch of 274,460 accounts, Twitter said 93 per cent were flagged by internal, proprietary tools, and 74 per cent were taken down before their first tweet.
Such stats are pointed towards governments in Europe, particularly the UK and Germany, which have consistently called for tech giants to tackle terrorist content faster and in a more automated way, threatening extra regulation and fines for those that don't play ball.
The UK Home Office went so far as to commission ASI Data Science to develop a tool to detect Daesh propaganda on YouTube – a move widely seen as a way of demonstrating to the public, and Silicon Valley, that it will take action if platforms won't.
However, Twitter pointed out that just 0.2 per cent of suspensions are now made following government reports, which is 50 per cent less than in the previous six months.
Twitter also argued the fact it took down 8.4 per cent fewer accounts in this period than the first half of the year demonstrates progress. The firm claimed that thanks to "years of hard work making our site an undesirable place for those seeking to promote terrorism, resulting in this type of activity increasingly shifting away from Twitter".
But the platform added its own counter-warning to governments, saying that regulatory actions risked damaging free speech, pointing to recent comments from Human Rights Watch about Germany's NetzDG law.
"The wave of regulatory pressure in Europe and beyond is setting an emerging precedent and creating a 'domino effect' as 'governments around the world increasingly look to restrict online speech by forcing social media companies to act as their censors'," the Twitter report said.
Giving itself a good old pat on the back for releasing its report, Twitter said that transparency is one of the main ways to ensure continued freedom of expression.
Meanwhile, governments remain happy to put in requests for information on Twitter users. For the latest period, governments made 3 per cent fewer requests than in the first half of 2017, but these specified about 52 per cent more accounts.
This included 1,158 emergency requests – where account-holders might not be told, for instance in a life threatening situation – about 1,549 accounts. Twitter gave some information in 54 per cent of such cases.
For non-emergency releases of information, governments made 6,268 requests about 16,861 accounts.
Twitter provided some information in 55 per cent of cases; of these, 93 per cent was non-content information, like email addresses or transactional data, while the rest was content, such as tweets or direct messages.
The most slurp-eager government was, as in previous reports, the US, which made 1,761 requests about 8,176 accounts. Twitter handed over some information in 77 per cent of cases.
This was followed by Japan, which made 1,518 requests and was successful in 67 per cent of cases. The UK government made 760 requests, France 365 and Germany, 237 – these countries were successful in 65 per cent, 65 per cent and 29 per cent of cases respectively.
In comparison, the Turkish government received no information for the 530 information requests it made about 2,583 accounts.
Twitter said that it could refuse to comply with requests if they were legally invalid or overly broad – although it said it will help the governments narrow down their requests. ®
Sponsored: Beyond the Data Frontier