Key facts
- Inman Grant told the Senate inquiry she had seen no indication over the past two years that companies are over-censoring.
- The Digital Duty of Care would require internet companies to proactively check and stop potential user harms.
- Companies that fail would face fines of up to A$109 million ($75 million).
- Inman Grant said most excessive content removal came from automated moderation without human oversight, not regulatory pressure.
- She said lawmakers should add a narrower definition of harmful content if they consider the wording too vague.
- The bill would also let users opt out of algorithmically recommended content and see only accounts they follow.
Australia's internet regulator, eSafety, on November 10 rejected warnings from technology companies that a proposed online safety law would lead to widespread censorship, telling a Senate inquiry that platforms' business models push them to maximize user engagement.
Commissioner Julie Inman Grant said she had not seen any indication, particularly over the past two years, that companies are over-censoring. "Companies today are trying to show the maximum amount of material, no matter how harmful or hurtful it is to the Australian community," she told the inquiry.
The government's Digital Duty of Care legislation would require internet companies to proactively check for and stop potential user harms, with fines of up to A$109 million ($75 million) for failures. The bill has passed the House of Representatives but needs crossbench support to clear the Senate and become law.
A senior executive at Google in Australia told the hearing on November 9 that a section requiring age checks where there is a risk of a child seeing pornography might force the company to block under-18s from Gmail. Representatives of Meta, which owns Facebook, Instagram and WhatsApp, as well as TikTok and Snap, raised similar concerns, saying the law could give regulators too much discretion to decide what is harmful.
Inman Grant said that if lawmakers considered the wording too vague they should include a narrower definition of harmful content, adding that more specificity would help enforcement. She said most cases of excessive content removal resulted from automated moderation systems operating without human oversight rather than from regulatory pressure.
The government wants the legislation, which would also let users opt out of algorithmically recommended content and see posts only from accounts they have chosen to follow, passed by the end of the year.