Agentic AI needs a purpose-drift register

Expansion of an AI system’s purpose, data access, or authority should create a visible governance event, says Dr Gleb Tsipursky of Disaster Avoidance Experts.

The UK is about to take an important step toward more concrete AI privacy governance. Regulations that entered into force in May(1) require the Information Commission (ICO) to prepare a statutory code of practice on personal data in artificial intelligence and automated decision-making, and the ICO is currently in drafting mode. The explanatory material says the code is intended to help organisations understand and comply with data protection law while supporting responsible innovation and growth.

That is a useful mandate. Yet the hardest privacy problem in agentic AI may emerge after an organisation has done its initial analysis correctly.

Continue Reading

UK Report subscribers, please login to access the full article

LOGIN

If you wish to subscribe, please see our subscription information.

Subscribe