Actors, musicians, crew members and journalists expressed concerns in a study released by the Media and Entertainment Arts Alliance on Wednesday, which also found many were unaware whether their work had been used to train generative AI models.

The findings come weeks before the federal government is expected to sit down with industry stakeholders and discuss the use of AI software to boost productivity at its economic roundtable.

Artificial intelligence experts warn the talks may not result in swift action on AI regulation, however, after policy changes in the United States and ongoing delays in formulating an AI law.

The media union surveyed more than 730 workers in creative industries, such as television, radio and film production, news media, art and music.

More than two in three (69 per cent) revealed they strongly agreed with calls for government intervention to regulate AI tools, and three in four (78 per cent) strongly agreed tech firms should pay for the work they used to train AI models.

Misinformation ranked as the top AI concern for respondents, followed by the loss of human creativity, the theft of work, and a lack of transparency about the technology.

The study highlighted serious and widespread concerns, MEAA chief executive Erin Madeley said, and followed a number of examples in which AI had been misused.

“We know that Australian voices, music and artwork have been scraped and faked, that ChatGPT is substituting the work of journalists, and that AI-generated clone hosts have been used for radio programs with no disclosure to audiences,” she said.

“This amounts to unsanctioned, unregulated and untaxed mining Australia’s creative resources.”

The study also found more than half of those surveyed did not know if their work had been used to train AI, and only three per cent had consented to its use and been compensated for their work.

While AI was expected to become a major focus at the government’s Economic Reform Roundtable in August, Ms Madeley said the talks should also centre on appropriate safeguards for employees.

“It is becoming increasingly clear that further government intervention will be required to ensure that productivity benefits arising from the use of AI filter down and are shared with Australian workers,” she said.

A Senate inquiry into adopting AI recommended a dedicated law to regulate the technology last year, and a consultation into mandatory AI guardrails attracted record submissions, UNSW AI Institute chief scientist Toby Walsh said.

Changing attitudes towards AI in the US and a change of minister had delayed AI regulations, he said, but the issue could not be ignored.

“There’s definitely a public appetite for it and when there’s a public appetite, politicians do move,” Professor Walsh said.

“There’s clearly significant public concerns around AI and the impacts it will have on jobs and different aspects of our lives so the pressure will surely be mounting on politicians to do something.”