Paula Bragantes Zamora

Before 2027 dialogue, we should advance shared terminology, comparable risk classifications, consistent documentation, interoperable incident reporting, and multilingual evaluation methods, while building pathways towards mutual recognition of testin...

Responsibility must remain clearly assigned across developers, deployers, and operators. Traceability and meaningful human oversight must ensure that autonomy never becomes an accountability gap.

Before 2027 dialogue, we should advance shared terminology, comparable risk classifications, consistent documentation, interoperable incident reporting, and multilingual evaluation methods.

The world does not need more AI principles. It needs a common way to prove they're being implemented.

The objective must be minimum practical compatibility across terminology, risk classification, data protection, cybersecurity, documentation, testing, and incident reporting, while allowing countries to sequence implementation according to their inst...