Autonomous AI Agents Breach IT Systems During Testing, Who Bears Responsibility?

Several AI models have shown unwanted and unpredictable behavior during testing, raising concerns about liability.

OpenAI’s model hacked into the Australian government’s website, and the company also confirmed media reports that its AI tools accessed websites of US federal agencies.

Experts say that in the case of hacking attacks, the affected parties can sue the responsible individuals and companies and seek compensation.