OpenAI calls for a pause in AI development

OpenAI, the developer of ChatGPT, has decided to slow down the pace of development of its most powerful AI systems. This decision was made in response to growing concerns about the risks of so-called frontier models. OpenAI intends to continue developing strong models only once clear and verifiable safety standards have been established. This step also impacts the European AI regulation and the requirements for governments and companies working with such systems.

The company has indicated that it will adhere to agreements with regulators, committing to better control the rollout of AI systems. As a result, the time for critical processes such as evaluations, red-teaming, and external audits will increase. It is expected that releases will be rolled out in a phased manner with stricter limitations, which will also lead to more stringent documentation requirements and API rules for developers.

Moreover, ChatGPT, like other generic systems, will be subject to the new AI regulation being developed in Europe. This includes a set of additional obligations for models that pose systemic risks. The legislation emphasizes that governments must ensure the safe use and procurement of such systems, taking into account privacy legislation such as the GDPR.

A slower development pace means that organizations in the Netherlands using AI applications like ChatGPT may expect fewer rapid upgrades. However, this may also result in more documented and predictable updates, which would be beneficial for risk analyses and tenders. Adjusting the development speed at OpenAI may ultimately contribute to a more manageable and safer AI environment where users benefit from increased safety requirements and transparency.

Read the full article from AI Insider.