OpenAI’s CEO, Sam Altman, took the stage with a whirlwind of change, displaying a remarkable transformation in his stance on European artificial intelligence (AI) regulation within a mere two days. First, he raised concerns about crossing a regulatory threshold and threatened to withdraw operations from Europe. However, he quickly backtracked, assuring everyone that OpenAI has no intentions of bidding farewell to the continent.
During an interview in London on Wednesday, Altman spoke openly to journalists, shedding light on his reservations regarding the European Union’s (EU) upcoming AI Act, which is slated for finalization in 2024. As reported by the Financial Times, Altman emphasized the significance of the fine print, stating, “The details really matter.” He went on to express his commitment to compliance but left no ambiguity about the potential consequence: “If we can’t comply, we will cease operating.”
Initially, the legislation had targeted “high-risk” applications of AI, such as in medical devices and decision-making processes related to hiring and loans. However, riding on the waves of the generative AI revolution, lawmakers have now proposed an expansion of the rules. Their gaze has turned toward creators of large-scale machine learning systems and tools, including powerful language models like OpenAI’s ChatGPT and Google’s Bard. These entities would be required to disclose AI-generated content and provide summaries of any copyrighted information utilized as training data for their systems.
OpenAI faced scrutiny for its lack of transparency regarding the methods and training data employed for GPT-4, one of the models that powers ChatGPT. The absence of such disclosures sparked criticism from various quarters.
Altman, in his London appearance, expressed his opinion on the current draft of the EU AI Act. He asserted that it veered towards over-regulation, but he also mentioned that he had heard rumors of potential amendments. “They are still talking about it,” he stated optimistically, as per Reuters. However, lawmakers quoted by Reuters countered that the draft was not open to debate. Dragos Tudorache, a member of the European Parliament from Romania, indicated a lack of belief in any significant dilution occurring in the near future.
Within a span of less than 48 hours from his initial warning, Altman took to Twitter, signaling a shift in his tone. He shared the news of a “very productive week of conversations in Europe about how to best regulate AI.” Altman’s tweet exuded enthusiasm as he affirmed OpenAI’s continued presence in Europe and quashed any notions of an impending departure.
As reported by the FT, the latest proposal for the EU’s AI Act will undergo negotiations between the European Commission and member states over the course of the upcoming year.