EU rules
for advanced AI are step in wrong direction, Google says
Google and
Meta are leading the charge against a “code of practice” governing tools like
Gemini, ChatGPT and Llama.
The European
Union is having to fight increasingly fierce criticism of its tech regulations
since U.S. President Donald Trump took office. |
February 10,
2025 4:28 am CET
By Pieter
Haeck
BRUSSELS — A
European Union plan to rein in the most advanced artificial intelligence models
is in peril after Google and Meta executives criticized the draft rules.
The EU has
been drafting a set of voluntary rules called a "code of practice"
that companies running the most advanced AI models (called general-purpose AI
models), such as OpenAI, Anthropic, Google, Meta and Microsoft, could sign up
to.
But the plan
is a "step in the wrong direction" at a time when Europe seeks to
restore its competitiveness, Kent Walker, Google's most senior public affairs
official, told POLITICO.
Walker's
comments added pressure, after Meta’s top lobbyist Joel Kaplan opened fire on
the code, saying the rules established “unworkable and technically unfeasible
requirements.”
The code of
practice is a follow-up to the EU's AI rulebook adopted last summer. The final
code is meant to give substance to what was said in the law. It touches on
thorny topics such as how to disclose which data was being used to train models
and how companies should deal with "systemic" risks.
Big Tech's
criticism falls just ahead of the AI Action Summit in Paris, where European
Commission President Ursula von der Leyen and tech boss Henna Virkkunen want to
show the EU is open for business.
The EU is
having to fight increasingly fierce criticism of its tech regulations since
United States President Donald Trump took office and backed American tech
giants in their message that EU laws and fines equate to "tariffs."
Meta's
Kaplan indicated to a Brussels audience in a video interview that the social
media giant wouldn’t sign the code in its current form. He said it went “beyond
the requirements” of the AI Act.
Walker said
that, for Google, it's "too soon to tell" whether the company will
sign the code, hinting that the AI summit in Paris could be a pivotal moment.
Google's
president of global affairs argued that the code threatens to introduce several
requirements, such as in copyright or third-party model testing, that go beyond
the scope of the exercise, were already addressed elsewhere, or put a burden on
the industry.
Work on the
code of practice is expected to wrap up in April, but its success hinges on
whether companies like Google and Meta sign up.
Sem comentários:
Enviar um comentário