Surprisingly, the European Union has allowed companies to audit themselves while the EU’s upcoming AI Act could have been stricter in its requirement. However, negotiators from the EU have recently struck a landmark deal on the world’s first comprehensive artificial intelligence (AI) rules. The newly agreed draft of the AI Act will require OpenAI, the company behind popular AI chatbot ChatGPT, and other companies to share key details about the process of building their AI products.
While the companies will still be auditing themselves, the AI Act is a promising development as AI companies continue to launch powerful AI systems with almost no surveillance from regulators. The law is slated to come into force in 2025 after EU member states approve it. It forces companies to shed more light on the development process of their powerful general purpose AI systems capable of generating images and texts.
According to a copy of the draft legislation, AI companies like OpenAI will have to share a detailed summary of their training data with EU regulators. This requirement aims to address the problem of biased data, which has resulted in troubling outputs from AI tools like ChatGPT that perpetuate sexist stereotypes.
However, the draft legislation could have gone further. It allows companies like OpenAI to hide certain key data points, including the kind of personal data used in their training sets. Additionally, AI companies can hide information about the prevalence of abusive or violent content and the number of content moderators they have hired to monitor their tools’ usage.
The AI Act is seen as a decent start in regulating AI, but critics argue it could have been more comprehensive. Nonetheless, it will be interesting to see if other regions, including the UK and the US, follow the EU’s lead and introduce similar regulations on AI in the future.
Overall, the AI Act’s requirement for increased transparency from AI companies is a step in the right direction. By compelling these companies to share key details about their AI models, researchers and regulators will be better positioned to identify and address potential issues with training data and biased outcomes. This move also highlights the need for continued progress in regulating AI technologies to ensure fairness, accountability, and transparency in their development and use.
Note: The EU’s AI Act does not mention OpenAI or any specific companies and is formulated in a general context.
