ipolletta’s commentary and the position paper are being distributed both at the national level in Italy and to

the European bloc—which, of course, on June 8 distributed a statement from the European parliament as it
works on what the EU is characterizing as “the world’s first comprehensive AI law.”

In Europe’s statement, we read, “Parliament’s priority is to make sure that AI systems used in the EU are
safe, transparent, traceable, non-discriminatory, and environmentally friendly. AI systems should be
overseen by people, rather than by automation, to prevent harmful outcomes.

“Parliament also wants to establish a technology-neutral, uniform definition for AI that could be applied to
future AI systems.”

One of the things helpful in almost any of the myriad early AI conversations and debates in world book
markets  are the EU’s starting points that define “unacceptable risk” and “high risk.”

In “unacceptable risks,” for example, the parliamentary process statement includes:

There’s a caveat there: “Some exceptions may be allowed: For instance, ‘post’-remote biometric
identification systems where identification occurs after a significant delay will be allowed to prosecute
serious crimes but only after court approval.”

In the category of “high risk,” the discussions include “AI systems that negatively affect safety or
fundamental rights.”

These are to be divided into two categories:

 “AI systems that are used in products falling under the EU’s product safety legislation. This includes
toys, aviation, cars, medical devices and lifts
 “AI systems falling into eight specific areas that will have to be registered in an EU database

In this area, the law is expected to require AI system assessments before those systems can be placed onto
the market “and also throughout their lifecycle.”

Under the heading of “generative AI”–which references the kind of “new” work that large language models
can make from “training”—the main concern the EU process lists is transparency:

 “Disclosing that the content was generated by AI

 “Designing the model to prevent it from generating illegal content
 “Publishing summaries of copyrighted data used for training”

The European Parliament’s process is timed to reach an agreement by the end of the year. Members of
parliament adopted the opening negotiating position on the AI act on June 14, activating talks with
member-states in the EC, the European Council, on the final form of the law.