In addition,
Europe tightens vice artificial intelligence:
The European regulatory machine is underway. In addition, For the past week, a first salvo of bonds from the artificial intelligence Act (AI Act) has entered into force. Consequently, The text. Consequently, adopted in 2023 and praised as the most ambitious in the world in the matter, intends to supervise the development and marketing of artificial intelligence on European territory. In addition, In sight: suppliers of so -called “general use” (or GPAI) models. In addition, capable of generating text or images, like Chatgpt (Openai), Gemini (Google) or Grok (XAI).
First rules for generative AI models
These companies must provide technical documentation on their models. Similarly, publish a summary of training data, and implement a copyright compliance policy. The models considered to be “at systemic risk”-that is to say presenting risks for security. fundamental freedoms or human control-will also have to notify the europe tightens vice artificial intelligence European Commission, assess the risks linked to their use, and strengthen their security.
The supervision is progressive: the models already in circulation have until August 2, 2027 to comply. But for the most powerful, the rules apply now.
And these new rules are not unanimous. The United States, and in particular the Trump administration, see this approach that they consider too restrictive. The American president even launched a “large ACA acceleration plan” to lift the regulatory brakes and strengthen American leadership.
On the business side, the reception is lukewarm. If Google has signed the code of good practices written by the Commission. the company also expressed its concerns, believing that this regulation risked ” brake the deployment of AI in Europe ». Meta, for her part, simply refused to sign the code. For Joel Kaplan, head of public affairs of the group, the text contains too many ” uncertainty ».
europe tightens vice artificial intelligence
However, this refusal does not exempt them from complying with the European regulation, which remains legally restrictive. The sanctions regime is ready: up to 7 % of global turnover for offenders.
Behind these technical discussions, it is a broader political confrontation that is played out. Brussels wants to assert its digital sovereignty and protect its democratic values. Washington denounces a ” inconsistent. restrictive approach “, In the words of Joel Kaplan, and threatens to include these regulations in future commercial negotiations.
Next step: the implementation of rules for high risk AI in 2026, in areas such as education, health or infrastructure.
🟣 To not miss any news on the Geek newspaper, subscribe to Google News and on our WhatsApp. And if you love us, we have a newsletter every morning.
Europe tightens vice artificial intelligence
Further reading: Customs duties: Trump announces that he has reached a commercial agreement with Indonesia – The state wants to tackle too numerous conventional ruptures. – “This underground treasure will change everything”: France amazed by the discovery of an immense rare and strategic gas deposit essential for the energy future – Orange victim of a cyber attack: what we know – “He pulverized all scores at 18”: Elon Musk’s mother publishes an unprecedented and scientific community test.