Facebook owner Meta says harmful content targeting First Nations during Voice referendum getting through

Meta has conceded its filtering tools on platforms like Facebook and Instagram could not be capturing the extent of hate speech and racism targeting First Nations people in the lead-up to the Voice referendum. Fronting a Senate committee on Tuesday, the social media giant said its tooling software to stamp out abusive and harmful content could be missing a number of comments directed at Indigenous communities online. Concerns were raised by Labor Senator Jana Stewart, who probed Meta about its monitoring of harmful content in the lead-up to the referendum and what the company was doing to ensure misinformation would not be amplified on its platforms. Senator Stewart also quizzed Meta about why content cannot be reported as “racism” on its platforms. Meta policy director Mia Garlick said the company had invested heavily in new technology to identify and remove hateful content which contravenes Australian racial discrimination laws. “In some instances, it can depend on how things have been expressed and it is quite possible that some of our tooling doesn’t capture that,” Ms Garlick said in relation to harmful content targeting Indigenous groups. “That is why we have developed new moderation tools to provide page owners in particular, or people for example who have Instagram accounts, the ability to . . . block certain words and to hide certain messages.” Ms Garlick added Meta was using a combination of site-wide and personal tools to assist people in combating information they do not want to see. She added further measures have been put in place to deal with content associated with the referendum being displayed on social media sites. “We do have an internal team that is stood up and trying to share information both that we’re receiving from external contacts, but also things that we’re identifying internally to see how we can continue to do everything we can to ensure that our policies and . . . all of our integrity measures are working as intended in the lead up to the referendum,” Ms Garlick said. Meta confirmed it had partnered with Reach Out in a specific campaign to address Indigenous mental health online. It said the decision to provide support in the space of mental health was due to the potential impact the Voice debate was expected to have on a number of communities. Meta’s accountability of its algorithms has been questioned in the past on its ability to spread harmful speech like wildfire. This was brought up during the hearing by Greens Senator David Shoebridge in relation to content being amplified on the violence and massacres targeting Rohingya Muslims in Myanmar in 2017. The content pushed through Facebook was designed to invoke more extremist views against Rohingya communities. Ms Garlick would not comment on the matter because it was already subject to current legal proceedings.