Vienna reviews every proposed use of artificial intelligence

Vienna is assessing each proposed use of artificial intelligence before approving it, the city’s chief information officer told ORF. Klemens Himpele said a team reviews cases from legal, technical and economic perspectives.

The city currently uses AI mainly to produce text and images, not in administrative procedures where residents submit applications for processing. Applications for parking permits, for example, are handled by conventional algorithms, Himpele said.

He said European Union rules allow AI to be used more readily where errors would have limited consequences or where a person remains in control. The rules are stricter for high-risk applications in which errors could have serious effects.

Himpele expects significant technological advances over the next one to three years, but said AI should be adopted only if it improves services for Vienna residents, for example by speeding up procedures or making dealings with public offices easier. Changes to Austria’s General Administrative Procedure Act have already created options for using AI in administrative processes, he added.

Himpele also said the city had found no indication that documents stolen in a recent cyberattack had been published. Attackers accessed city IT systems and stole about 26,000 internal documents, including data relating to 6,000 people.

The city faces ongoing cyberattacks, with waves becoming stronger, Himpele said. He said AI has made attackers’ automated searches for vulnerabilities more effective, prompting the city and software companies to use similar tools to find and fix weaknesses.

Himpele said the attackers in the recent incident did not take over user identities or gain control of the city’s systems. Vienna’s administration operates about 10,000 servers and stores 47 petabytes of data, he said.