OpenAI warns of break with Europe as regulation progresses

OpenAI chief Sam Altman has warned that Brussels’ efforts to regulate artificial intelligence could prompt the maker of ChatGPT to pull its services out of the EU, in the starkest sign yet of a growing transatlantic divide over the technology’s governance.

Speaking to reporters during a visit to London this week, Altman said he has “many concerns” about the EU’s proposed AI law, which is due to be finalized next year. In particular, he pointed to the European Parliament’s move this month to extend the proposed regulation to the latest wave of general-purpose AI technology, including large language models such as OpenAI’s GPT-4.

“The details really matter,” Altman said. “We will try to comply, but if we cannot comply, we will cease operations.”

Altman warned U.S. tech companies are bracing for what some predict will be a protracted battle with European regulators over a technology that has shaken up the industry this year. Google CEO Sundar Pichai also toured European capitals this week to influence policymakers as “safeguards” are drawn up to regulate AI.

The EU’s AI law was originally designed to address specific, high-risk uses of AI, such as in regulated products such as medical equipment, or when companies use it to make important decisions, including lending and hiring decisions.

However, the sensation caused by the launch of ChatGPT at the end of last year prompted a rethink, this month the European Parliament defined extra rules for widely used systems, which are of general application beyond the previously targeted cases. The proposal still needs to be discussed with the member states and the European Commission before the law enters into force by 2025.

The latest plan calls for creators of “core models” — the large systems behind services like ChatGPT — to identify and try to mitigate the risks their technology could pose in a variety of environments. The new requirement would make the companies that develop the models, including OpenAI and Google, partially responsible for the use of their AI systems, even if they have no control over the technology’s embedded application.

The latest rules would also force tech companies to publish summaries of the copyrighted data they used to train their AI models, so artists and others could try to claim compensation for using their material.

The attempt to regulate generative artificial intelligence, while the technology is still in its infancy, “showed fear on the part of lawmakers who read the news like everyone else,” said Christian Borggreen, European head of the Washington-based Computer and Communications Industry Association. . US tech companies supported the EU’s previous plan to regulate AI before the “knee-jerk” reaction to ChatGPT, he added.

US technology companies have urged Brussels to be more cautious in regulating the latest artificial intelligence, arguing that Europe should wait longer to study the technology and work out the balance of opportunities and risks.

Pichai met with officials in Brussels on Wednesday to discuss AI policy, including Brando Benifei and Dragoş Tudorache, the lead MEPs responsible for the AI ​​law. Pichai stressed the need for proper regulation of the technology that didn’t stifle innovation, according to three people present at those meetings.

Pichai also met with Thierry Breton, the EU digital chief who oversees the AI ​​law. Breton told the Financial Times that they had discussed introducing an “AI Pact” – an informal set of guidelines for AI companies to adhere to before formal rules come into force – as “there was no time to waste in the AI ​​race for a for building a safe. online environment”.

US critics say the EU’s AI law imposes sweeping new obligations to control risks from the latest AI systems without setting specific standards they are expected to meet.

While it’s too early to predict the practical effects, the law’s open-endedness could prompt some U.S. tech companies to rethink their European presence, said Peter Schwartz, vice president of strategic planning at software company Salesforce.

He added that Brussels will act “without reference to reality, as it has in the past” and that with no European company leading the way in advanced artificial intelligence, the bloc’s politicians have little incentive to support the industry’s growth. “Essentially, European regulators will regulate American companies, as they did in the IT era.”

The European proposals would prove workable if they “continue to require companies to keep up with the latest research [on AI safety] and the need to continuously identify and mitigate risks,” said Alex Engler of the Brookings Institution in Washington. “Some of the uncertainties can be filled in later by the EC and standards bodies.”

While the law appeared to target only large systems like ChatGPT and Google’s Bard chatbot, it risks affecting “open-source models and non-profit uses” of the latest artificial intelligence, Engler said.

Executives from OpenAI and Google have said in recent days that they support the eventual regulation of AI, though they have called for further investigation and debate.

Kent Walker, Google’s President of Global Affairs, said in a blog post last week in which the company supported efforts to set standards and reach a broad political agreement on artificial intelligence, such as those underway in the US, UK and Singapore – while pointedly avoiding comment on the EU , which is the most advanced in the adoption of artificial intelligence. specific rules.

The political timeline means Brussels could choose to move forward with its current proposal rather than trying to develop more specific rules as generative artificial intelligence advances, Engler said. A longer time to refine the artificial intelligence law would risk delaying it during the current EU presidency, which could push the whole plan back to the drawing board, he added.