A survey in Kaohsiung found that older residents identified AI-generated material correctly only about 28 percent of the time, the Liberty Times reported. Two city council candidates proposed cross-sector cooperation on AI education for seniors.

As voices, images and text produced by generative AI enter everyday channels of communication, judging authenticity has ceased to be one generation's problem. The survey nonetheless shows the difficulty is pronounced among older people.

The practical risk is voice phishing and investment fraud. Methods using synthesised voices to impersonate family members, or composited video of well-known figures to solicit investment, have been reported in many places.

Three responses are commonly cited: repeated education at neighbourhood level, delay procedures on transfers at financial institutions, and labelling obligations for synthetic material on platforms. The shared observation is that any one of them alone has limited effect.

Comparable surveys and responses continue in Korea and Japan. Societies with a high share of older residents face similar conditions, since the social cost of the problem grows with that share.

In communities that use several languages together, such as the Chinese diaspora, the language of educational material must also be considered, since guidance that is not translated reaches few people.