PUBLISHED ONMarch 23, 2026 9:45 AMBYDana LeongMisleading advertisements created with generative artificial intelligence is on the rise and ethical guidelines on AI usage is needed, according to data collected by the Advertising Standards Authority of Singapore (ASAS) in 2025.
In a press release on Monday (March 23), the advertising watchdog, which is an advisory council to the Consumers Association of Singapore, said that it received a total of 379 pieces of advertisement feedback last year.
Of all the feedback collected, there were seven concerned advertisements that involved the use of generative AI — more than double the number received in 2023 and 2024 combined.
ASAS found that most concerns regarding generative AI usage in advertising relates to the creation of misleading advertisements.
"The most common concern was that the images and videos in the advertisements presented fictitious events or misleading claims," the watchdog said, adding that it received feedback from various industries spanning from interior design to investments.
One such example involved a telemedicine company that used generative AI to create a video where a male character gave a first-person testimonial about their services.
"Complainants said that the advertisement gave the impression that it came from an actual individual when this was not the case," said ASAS, adding that the advertiser eventually ceased the advertisement and conducted an internal review to ensure it would not recur.
The watchdog said that the use of AI to create advertising content is "not inherently objectionable", but that responsible use for honest decent and legal ethical advertising is paramount.
"False and misleading advertisement content is unacceptable, regardless of the tools used or the disclosures added," the advertising watchdog said.
As such, sector-specific guidance on AI use is necessary to protect consumers and stakeholders, and ASAS suggests a risk-based approach that focuses on labelling requirements and usage restrictions on applications where generative AI usage has high potential to mislead.
[[nid:731919]]
