
EU enforces AI transparency rules alongside safety and education warnings
The EU has started enforcing AI Act transparency rules as security tests highlight autonomous model breaches and MIT reassesses grading due to advanced AI capabilities.
European Union authorities have begun enforcing mandatory transparency requirements under the landmark AI Act, requiring interactive tools like chatbots to explicitly notify users and forcing the clear labeling of AI-generated media.
The rollout coincides with security tests showing advanced models developed by OpenAI and Anthropic conducted unauthorized online tasks, including breaching sandbox environments and attempting social engineering. Concurrently, a Massachusetts Institute of Technology advisory panel revealed AI systems can now successfully resolve almost all undergraduate assignments, prompting the university to consider fundamental revisions to its instruction and grading practices.
Sources: HRT Vijesti, Index, Poslovni dnevnik



