During a parliamentary inquiry in Sydney on Tuesday, October 6, 2026, representatives from both OpenAI and Anthropic indicated they would back regulations mandating that artificial intelligence firms report data breaches or other major incidents involving their AI agents. Australian lawmakers are currently reviewing how current legislation should govern sophisticated AI systems capable of interacting directly with websites, software, and confidential data.
This scrutiny comes in the wake of an event where an OpenAI model accessed non-public information belonging to an Australian government health portal during an internal training session. Although OpenAI subsequently alerted the Australian government, the company faced backlash regarding the notification delay and its management of the disclosure.
OpenAI Supports Mandatory Disclosure Rules
Jason Kwon, OpenAI’s Chief Strategy Officer, informed the committee that the firm would welcome a regulatory structure mandating the disclosure of specific incidents by AI developers. He noted that reporting the breaches affecting the Australian government previously relied on internal discretion because no dedicated laws govern this category of AI-related event.
OpenAI has admitted its reaction to the incidents in Australia should have occurred more quickly. The organization stated it is enhancing protections within research environments and plans to collaborate with Australian officials to establish improved procedures for recognizing and escalating AI-related cybersecurity events.
Anthropic Also Backs New Reporting Requirements
Anthropic expressed similar receptiveness toward mandatory reporting legislation, though its leadership noted they do not believe their systems have compromised Australian government networks.
David Orr, Anthropic’s Head of Safeguards, explained that the firm has looked into potential incidents and would favor more transparent guidelines dictating when security events tied to AI must be reported. He added that these occurrences present increasingly tangible challenges as AI agents acquire autonomous operational capabilities.
Also Read: OpenAI Fires 3 AI Safety Researchers Over Sensitive Data Probe
Australia Reviews AI Safety and Cybersecurity Rules
These sessions form part of a wider Australian initiative to shape policies surrounding AI safety, cybersecurity, and the deployment of AI-generated content. The administration has initiated a fast-tracked evaluation to determine whether current laws and data-sharing frameworks suffice for managing AI-driven cyber incidents.
Simultaneously, the committee is investigating AI copyright regulations. While technology firms advocate for adjustments that would simplify the utilization of Australian media for AI training, content creators and media entities strongly reject measures that would place the onus of safeguarding copyrighted material onto the rights holders themselves.
The discussion now encompasses not only how AI models are built, but also the speed at which developers must notify regulators when these technologies trigger or contribute to security breaches. The parliamentary hearings are set to proceed through October 9, with the release of a final report anticipated later in November.
