Altman took part in a panel discussion at University College London, during which he said the ChatGPT creator may have to pull out of Europe owing to what he perceives as impossible regulations.
“If we can comply, we will, and if we can’t, we’ll cease operating… We will try. But there are technical limits to what’s possible,” he said, according to TIME.
This key issue is OpenAI’s popular large language models (LLMs) may be labelled “high-risk” by the EU’s regulatory framework, which is currently being formulated.
High-risk AI entities include those that have a hand in infrastructure, certifying product safety, validating credit scores, analysis of travel documents, scoring exams and more.
GPT-4 is a huge AI model based on both text and image data, trained using a rumoured 170 trillion parameters. It could theoretically be used in all these situations, given the right application of the technology.
If OpenAI’s LLMs are labelled as “high-risk”, much more stringent rules will apply to them. According to the European Commission website, this will include oversight of the quality of the data used to train the models, and a “logging of activity to ensure traceability of results.”
Both of these seem incompatible with GPT-4 given its absolutely huge training dataset and near-uncontrolled applications. It could also be argued current forms of AI fail the EU’s demand for “high level of robustness, security and accuracy.”
Altman appeared to shrug off the potential for LLMs to be used as factories for disinformation and misinformation, suggesting the onus should instead rest on the social media platforms through which it is typically distributed.
“You can generate all the disinformation you want with GPT-4, but if it’s not being spread, it’s not going to do much,” Altman said.
Is AI Dangerous?
Italy’s Data Protection Authority banned ChatGPT, the famous chatbot based on GPT-3.5, in March 2023 over privacy concerns. However, the ban was lifted at the end of April after some changes were made to operating practices.
In a March 2023 interview with ABC News, Altman admitted that ChatGPT could “eliminate” a lot of jobs, but could in turn lead to “much better ones.” He did not have any suggestions as to what these may be.
The UK government has to date said it has a “pro-innovation” stance on AI. However, PM Riski Sunk met with AI leaders in OpenAI, Google DeepMind and Anthropic AI to discuss how to proceed yesterday evening, May 24.
“They discussed safety measures, voluntary actions that labs are considering to manage the risks, and the possible avenues for international collaboration on AI safety and regulation. The lab leaders agreed to work with the UK Government to ensure our approach responds to the speed of innovations in this technology both in the UK and around the globe,” reads a statement released by the government.