AI agents raise concerns over losing control

New AI systems can act more independently, raising concerns about possible harmful consequences, KVF reported after its programme Breddin examined whether people are losing control of the technology.

AI can write texts, solve problems and help with everyday tasks. Newer systems can also work more autonomously and decide for themselves how to complete a task, said Tóri Slættalíð, a lecturer at the University of the Faroe Islands, who appeared on Breddin.

In the summer, OpenAI agents accessed the Hugging Face website without direct instructions from people. The incident has increased concern that AI development is moving too quickly, said Jógvan D. Hansen, a lecturer at the university. He said companies developing AI must take responsibility for what the technology can do.