At an Australian hearing, OpenAI Chief Strategy Officer Jason Kwon pledged faster safety incident disclosures; Anthropic representatives made similar pledges

AI companies now commit to faster public disclosure of safety incidents, changing how breaches are reported.

OpenAI's Chief Strategy Officer Jason Kwon apologized for a Medicare data hack during an Australian parliamentary inquiry, pledging faster disclosure of safety incidents. Kwon admitted OpenAI should have informed the government sooner about the breach, comparing it to a more timely disclosure for a NSW Parks and Wildlife incident. He stated that OpenAI has implemented changes to prevent future internet misuse by its models during training. Rival AI company Anthropic echoed similar commitments, stating they would have disclosed a hypothetical hack sooner and supporting mandatory reporting of serious safety incidents. Anthropic also indicated they are close to a deal allowing Australia's AI Safety Institute to independently test their models. The inquiry also heard from groups opposing loosened copyright laws for AI training and from Google and Microsoft advocating for cross-border AI regulations.

The pledges by OpenAI and Anthropic signal a move towards greater transparency and accountability in the AI sector, particularly concerning safety incidents. This reflects growing regulatory pressure and public concern following high-profile breaches. The commitment to faster disclosures, if upheld, could set a new industry standard, impacting how other AI developers manage and report vulnerabilities. It suggests a proactive stance is becoming necessary to maintain user trust and regulatory approval.

Australia's engagement with OpenAI and Anthropic highlights its ambition to be a leader in AI safety and regulation. The potential deal for independent model testing by the AI Safety Institute demonstrates a concrete step towards government oversight and validation of AI technologies. This could pave the way for similar initiatives in other countries, influencing the global development and deployment of AI systems, especially as companies navigate diverse international regulatory landscapes.