OpenAI, Google and Microsoft outline doomsday scenarios about AI

Major tech companies, including OpenAI, Google, and Microsoft, emphasize the growing concern over potentially harmful consequences of artificial intelligence (AI). Despite these warnings, they continue to develop advanced systems, such as ChatGPT and Gemini, at a fast pace. This leads to criticism in Europe, particularly in the Netherlands, that if the technology is so risky, it might be better to roll it out slowly or not at all. The timing of this discussion is relevant, as the European AI regulation (AI Act) is about to be implemented, forcing governments to make clear choices.

Companies like OpenAI and Google acknowledge the risks of AI systems, such as misuse and misinformation, but continue to expand and improve their products globally. Critics point to the contradiction between the warnings and the ongoing launch of new technologies. Many advocate for clear standards and controls on the rollout of AI systems, while companies claim they implement sufficient safety measures and testing. However, concerning transparency, much information about the testing processes remains behind closed doors, making it difficult to verify these companies' claims.

The upcoming AI regulation establishes a risk-oriented approach that must regulate newly developed AI systems. The higher the risk of a system, the stricter the requirements are for developers and users. This means that generic models like ChatGPT and Gemini must adhere to stricter transparency and quality rules, potentially with additional oversight from the European AI Office. For organizations in the Netherlands, this means that procurement contracts must comply with the new regulations, with requirements such as logging and explainability within AI applications.

With the implementation of the AI regulation, the responsibility lies with companies to act proactively. It is now up to them to establish clear “no-go” criteria to ensure that AI models meet safety requirements before they are launched. This allows organizations to strengthen their own risk management by including audit agreements and incident reporting in contracts. In the coming months, during the rollout of the AI regulation, it will become clear whether tech companies are truly delivering on their promises.

Read the full article from AI Insider.