In coming weeks the British government is expected to release a white paper that may propose an internet safety commissioner or social media regulator, whose remit will cover not only violent and extremist content but also online child sex abuse and fake news.
It’s speculated that Britain may be preparing to legislate a “duty of care” and code of conduct for the technology industry, creating legal obligations to proactively filter out harmful content.
These proposals fit a broader pattern of tough talking by Europe, which often acts tough on the tech giants as well. There are proposals to widen the tax net on the firms, and also a stream of court cases and regulation on protecting user privacy.
Europeans were enthusiastic supporters of a previous G20 push on e-terrorism, which focused on industry self-regulation. The communique issued after the Hamburg meeting of G20 leaders in 2017 called on industry to “invest in technology and human capital to aid in the detection as well as swift and permanent removal of terrorist content”.
But patience has since worn thin, and governments are increasingly beating the drum for a tougher, regulation-based approach.
Progress on developing and implementing take-down laws is slowed, though, by the complex political and practical issues.
The big tech companies, and some civil liberties organisations, have argued that the industry is being asked to do the state’s job, deciding what is harmful or hateful and what isn’t. The companies say they’re uncomfortable at having to take on that level of risk and responsibility.
There’s also the argument that because hate speech is tricky to define, it’s hard to set regulations that don’t run the risk of trampling on free speech. One concern is that companies will take a risk-averse approach in response to the threat of big fines, and so will block far more content than they should.
And then there’s the potential cost: unless and until machine learning can readily and reliably identify terrorist or extremist content, a large amount of manpower has to be devoted to what is potentially a huge volume of content. That has to come from either the stretched public sector or the bottom-line-focused private sector.
Finally, there’s the question of how effective the law will be: content not only needs to be removed, it needs to be removed fast. The vast majority of hits on a piece of terrorist propaganda take place in the first few hours after it is posted. So if it takes longer than that to get the post down, it’s hardly worth the effort.
Germany took on-board these issues and ploughed ahead with its law, known as ‘NetzDG’, which came into force in January last year.
The law creates a process by which people can complain to the companies about hate speech online. A study by the Europe-based Counter Extremism Project last November found that Twitter and Google were rejecting around 80 per cent of complaints.
The think tank also found that Facebook had deterred complaints by making the NetzDG law’s complaints-reporting process difficult to use. But the company was unlikely to be fined heavily, because it could point to its own self-policed standards: while it had taken down just 74 pieces of online content in response to NetzDG complaints during the law’s first six months, worldwide over the same period it culled 6 million pieces of content for breaching its community standards.
The report found other flaws in the law. Extremist videos can be re-uploaded, and this was happening with 91 per cent of ISIS videos. But the tech companies had won a campaign to ensure NetzDG didn’t capture re-uploads, so that each re-upload must be treated as an entirely new case.
The report also said smaller platforms would find the NetzDG red tape much more difficult to cope with than the big three social media companies.
Britain has been fretting, during the process of producing its white paper, that too heavy-handed an approach might stifle its burgeoning tech sector.
The European Commission wants to prevent countries around the Continent from creating a patchwork of potentially inconsistent laws. So last September it put forward a law to combat terrorism and political extremism online.
The proposed law’s ambit extends beyond social media platforms to messaging apps and cloud storage providers, and would require these companies to take down content flagged as terrorist or extremist within an hour. To avoid the risk of companies being over-zealous in blocking material, the definition of terrorism is limited to an agreed list of terrorist groups.
A key challenge with taking this agenda to the G20 is that not every member is a western liberal democracy. China, for example, could use such a debate to push for a much broader definition of what constitutes harmful speech – in effect using a proposal aimed at terrorism to justify further domestic political repression.