OpenAI and Anthropic back data-breach reporting laws in Australia

Executives from Anthropic and OpenAI told an Australian parliamentary inquiry on Tuesday that they do not believe their products have breached Australian government systems, but that they would be open to laws that would force artificial intelligence companies to disclose data breaches, as Reuters reports.

The statements come several weeks after, as Reuters reported, OpenAI revealed that one of its agents had penetrated the country’s main health portal, Medicare — an event for which Australia criticised the American company in September, after it had notified the government around three months after the June incident.

Anthropic’s head of security, David Orr, told the committee that a “long and thorough investigation” following the penetration by OpenAI’s agent had found no breach of Australian government systems. “We have found nothing of the kind, and we have looked,” he said, according to Reuters. He added that the company had “much less visibility” over what had happened at its clients, as they usually had stricter rules on deleting data.

Anthropic’s head of policy for Australia and New Zealand, David Masters, told investigators that the company would be open to Australian laws requiring artificial intelligence companies to disclose data breaches. According to the Australian news agency AAP, Anthropic’s special envoy and former U.S. ambassador to Australia, Jeff Bleich, acknowledged that “for now this is largely voluntary”.

Australia is preparing specific artificial intelligence laws that will enter into force next year, while both companies — as Reuters reports — are also asking Australia to ease its copyright laws for the use of domestic content in training models, and are awaiting approval for massive data centres planned in Sydney and in regional Queensland.