CANBERRA — Representatives from OpenAI and Anthropic appeared before an Australian Senate committee on Tuesday, defending their data practices and safety records as lawmakers pressed for binding regulation of frontier AI systems.

The Senate Select Committee on Adopting Artificial Intelligence, chaired by Labor senator Tony Sheldon, had summoned the US firms after earlier hearings examined the use of Australian copyrighted material to train large language models.

Senators questioned the companies over the sourcing of training data, protections for minors using chatbots, and the concentration of market power among a handful of American developers. Representatives from both firms defended their voluntary safety commitments while resisting calls for a dedicated AI Act modelled on European Union rules.

The hearing intensified pressure on the Albanese government, which has signalled it favours mandatory guardrails for high-risk AI applications rather than relying on existing law. Senator Sheldon has repeatedly argued that self-regulation by technology companies has failed to protect Australian creators and consumers.

"These companies are building products on the back of Australian work without permission or payment," Sheldon said during the session.

The committee is due to report its final recommendations later this year. Industry groups including the Media, Entertainment and Arts Alliance have urged the government to legislate licensing requirements for AI firms using local content.