Three weeks after a gunman livestreamed to Facebook as he killed 50 people in Christchurch, multiple copies of the horrific clip remain on the social network.
The Herald asked Eric Feinberg, founder of the New York-based Global Intellectual Property Enforcement Center (GIPEC), an organisation that tracks extremist content on social media, to see if he could locate any copies of the clip.
This morning, the researcher told the Herald he had found seven examples of the gunman's video still live on Facebook, and five on the Facebook-owned Instagram.
"We continue to work around the clock to remove any new uploads of re-edited violating content using a combination of technology and people."
Feinberg located "a handful" of examples of the gunman's clip on Twitter and Google-owned YouTube in the immediate wake of the Christchurch attacks, but both services were clear today.
"Social media companies are making a great effort to eradicate extremist content. But based on what our systems find on a daily basis they have a long way to go," Feinberg said.
In a March 30 open letter, Facebook COO Sheryl Sandberg said, "We are committed to reviewing what happened and have been working closely with the New Zealand Police to support their response."
"We are exploring restrictions on who can go Live depending on factors such as prior Community Standard violations," Sandberg wrote.
"We are also investing in research to build better technology to quickly identify edited versions of violent videos and images and prevent people from re-sharing these versions."
For NZ Privacy Commissioner John Edwards, considering restrictions on some users was not good enough.
Edwards - who maintains Facebook should have added more safeguards to Facebook Live before launching what he sees as a "high risk" service in 2015 - asked the social network early this week to answer the question, "Is there anything that Facebook has done in relation to the livestreaming service since March 15, which, if it had done before March 15 would have either prevented the livestream occuring, or ensured it could have been more promptly deleted and either halted, or referred to a human for review?"
Facebook's answer was "no," Edwards says.
AI safeguards failed
In an interview with the Herald on March 19, Facebook global policy VP Monika Bickert told the Herald that around 200 people watched the gunman's 17-minute video live.
Facebook's artificial intelligence (AI) systems failed to notice the clip. It was taken down after around an hour, after NZ law enforcement alerted the social network to its presence.
Around 4000 people viewed the original video before it was taken down, Bickhert said. Some took copies, which turned up on Twitter and YouTube.
In the 24 hours after the attack alone, there were 1.8 million attempts to upload the video. Some 1.5m were blocked by Facebook, but another 300,000 initially slipped through.
Bickhert said Facebook staff were working around the clock to block the clip and stamp out copies.
She said AI systems were getting better all the time, but still imperfect.
Aussies crack down
Yesterday, Australia's Parliament passed a law that provides for fines of up to 10 per cent of social network's revenue or up to three years' jail for its executives if it is not swift enough in taking down "abhorrent" content.
Here, in notable contrast to its moves on gun-control, our government has yet to outline or timeline any measures to crack down on social media.
The Christchurch gunman's video and his so-called manifesto have been rated objectionable by chief censor David Shanks, meaning it is illegal to possess, view or share them. Individuals facing a fine of up to $10,000 or up to 14 years' jail if they ignore Shanks' ruling. A teen and a Christchurch businessman were both held in custody after sharing the clip.