Google revealed plans to tap into more of its users' personal information on Wednesday, as part of a raft of AI announcements at its annual developers conference, Google I/O.
But company executives stressed private data would only be accessed with owners' permission and that consent could be withdrawn.
The announcements come days after the Productivity Commission named AI regulation as one of its targets for urgent reform as the federal government had yet to introduce mandatory guardrails for its use.
Google announced several AI advances at its event in Mountain View, California, and chief executive Sundar Pichai said consumer and business demand for AI tools had grown exponentially.
"We're in a new phase of the AI platform shift where decades of research are now becoming a reality for people, businesses and communities," he said.
"More intelligence is available for everyone, everywhere, and the world is responding, adopting AI faster than ever before."
The number of data tokens processed by Google AI tools had grown by 50 times in the past year, Mr Pichai said, and were expected to grow further with improved models and features.
AI tools announced at the conference included one that would personalise responses for users by tapping into their information stored in other Google apps.
"One example we are introducing is personalised smart replies in Gmail," Mr Pichai said.
"It takes smart reply a step further, pulling relevant information from notes and Google Docs and past emails, matching your tone and style and generating a comprehensive email."
Google also announced a personalised AI search tool called AI Mode that could connect to Gmail accounts to share personal information, such as a person's restaurant or movie preferences.
Allowing AI to access this information would deliver more useful results, Google Labs and Gemini vice president Josh Woodward said, although the features would be offered on a strict opt-in basis.
"You, as a user, will be in control, it will be transparent about what information is being shared, and it will give you the ability to turn it on and off," he said.
"This is a way we think it can provide more help and be way more useful for people."
Other announcements at the event included AI-generated audio for background sound or dialogue in videos, DeepThink for advanced maths and coding output, and an extension of the company's "virtual try-on tool" that would artificially apply clothes to a user's own photo.
AI use has grown in Australia with half of all adults regularly using the technology, according to a study by the University of Melbourne and KPMG, but only one in three trust its results.
The federal government launched voluntary AI guidelines in 2024, though it has yet to launch mandatory guardrails for high-risk uses, which the Productivity Commission named as a priority on Monday.
