A decision by a Queensland government department not to disclose the use of artificial intelligence (AI) to create social media posts has prompted warnings about the importance of transparency as technology develops.
An illustration featuring a floating fishing rod and gibberish text was posted to the Queensland Fisheries Department’s social media accounts warning about relying on information provided by AI.
The caption read, “Don’t trust AI with fishing rules.”
AI-generated image from Queensland Fisheries social media pages. (Provided by: Queensland Fisheries Department)
Curtin University internet studies professor Tama Lieber said the post was “ironic”.
“It was an AI-generated post that said, ‘Don’t trust AI,'” he said.
The ABC discovered late last year that at least four posts from Queensland Fisheries’ Instagram and Facebook pages were created using an AI image generator.
The department typically shares content about protecting the state’s fishing resources to an audience of more than 143,000 followers across its Instagram and Facebook accounts.
Curtin University internet research professor Tama Lieber said it would become increasingly difficult to tell what is real and what is AI. (ABC News: Keene Burke)
Posts featuring AI-generated images discussed issued infringement notices, patrol operations, and recent court cases.
When tested with the Synth ID detector, two images returned positive results for Google’s AI tool’s invisible watermark.
The detector said it found that the other images showed several visual signs that “strongly suggested” they were generated by another AI tool.
None of the four posts disclosed in the caption or alt text that they were created with an image generator.
An AI-generated image posted by Queensland Fisheries warning of crab pot theft. (Provided by: Queensland Fisheries Department)
Lieber, who is also a principal scientist at ARC’s Center of Excellence for the Digital Child, said it would be “a good thing” for governments and the public sector to allow the use of AI, as it “becomes increasingly difficult to tell” when AI is being used.
He said it was already “very easy” to create cartoonish, figurative and photorealistic images using the technology.
“We are in a period of transition. Best practice in the ideal sense is complete transparency of how AI is being used to generate images.”
River said.
“We’re not at the point where it’s a legal requirement.”
“We have not received any concerns.”
A spokesperson for the Department of Primary Industries, which manages social media accounts for the Queensland Fisheries Department, confirmed that AI was used to generate the channel’s images.
They said the content was created “for illustrative purposes, particularly where privacy, legal, and operational considerations preclude the use of actual images.”
One of the social media posts created using Queensland Fisheries AI. (Provided by: Queensland Fisheries Department)
“We have not received any concerns that images used on our social media channels are blurred or mistaken for real images rather than AI-generated illustrations,” a spokesperson said in a statement.
Guidelines for the use of generative AI for Queensland government agencies say the technology offers an opportunity to improve productivity, but recommend that employees ensure that content generated by AI is “clearly” identified.
Primary Industries Minister Tony Perrett has also been contacted for comment.
During the 2024 state elections, the currently elected LNP government circulated an AI video of Labor leaders dancing on TikTok.
At the time, the party defended the use of deepfake videos, saying they were “clearly shown to have been created by AI”.
public expectations
Marketing professor Paul Harrison said more government agencies across Australia were using AI for social media content.
“They are using it because, like every other organization and sector, they are under pressure to become more efficient,” the Deakin University academic said.
Marketing professor Dr. Paul Harrison says posts should be labeled as being generated by AI. (Supplied)
Dr Harrison said the public does expect government agencies to “act appropriately and be transparent”.
He said the images on Queensland Fisheries social media accounts were “clearly AI-generated” and “we should say so”.
Dr Harrison said that the moment people learn that something was created by AI, they have a negative reaction to it.
But he said not disclosing the technology also created “another risk” of people asking “why didn’t you tell me it was AI?”
“If you don’t feel comfortable disclosing it, you should ask, ‘Why are you feeling comfortable disclosing it?'” Dr. Harrison said.
He said that from a marketing perspective, AI-generated images on social media are “not very good.”
”Was this the most effective way to get people’s attention?”
“At the end of the day, to me this looks more like laziness than anything else,” Dr. Harrison said.
