Updated Guidelines for AI: Alterations for ChatGPT and Similar Services
The European Union (EU) has introduced a landmark regulation, the AI Act, which aims to ensure transparency and ethical use of Artificial Intelligence (AI) technologies. The Act will come into force on August 2, 2025, affecting millions of Europeans who use AI tools daily[5].
Under the AI Act, providers of General Purpose AI (GPAI) models, including advanced chatbots and language generators, must comply with detailed transparency and copyright obligations when placing these models on the EU market[1][3]. This includes providing clear information about how the AI models are trained, the capabilities and limitations of the AI system, and the AI system’s purpose and potential risks[1].
For highly advanced or systemic risk models (exceeding set computational thresholds), additional transparency obligations include notifying the European Commission and ensuring continuous safety and security measures[1]. AI systems intended to manipulate human behaviour subliminally or exploit vulnerable populations are expressly prohibited[2][3].
Providers must also publicly document which copyright-protected contents they used for training, which is relevant for content creators whose works may have been used without consent[3]. Images, videos, and texts generated with AI must be recognizable as artificial through technical solutions like watermarks or metadata[3].
Social networks like Facebook, Instagram, YouTube, and LinkedIn are increasingly clearly labeling AI-generated contents to avoid misunderstandings. However, the labeling requirement cannot completely prevent deepfakes due to AI models from non-EU countries[4].
Not all AI providers are fully on board with the AI Act. Meta, the corporation behind Facebook and Instagram, has categorically rejected the AI Act's code of conduct and criticizes it for going beyond the actual law[6]. OpenAI, the company behind ChatGPT, had initially threatened a complete EU exit but quickly backed down[6]. Anthropic, the provider of the ChatGPT competitor Claude, shows more cooperation and wants to sign the code of conduct, signaling a willingness to collaborate[6].
Users of AI models will gradually notice changes, as providers must ensure that users can clearly recognize if they are communicating with an AI system[4]. Those who create AI-generated avatars of real people or use AI texts to inform the public must clearly label this[4].
The AI Act positions Europe as a pioneer in regulating Artificial Intelligence, setting a global standard for transparency and ethical AI use[1]. Each EU member state must designate a market surveillance authority by August 2 to effectively enforce the new AI rules[1].
References:
- European Commission. (2022). Proposal for a Regulation of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Retrieved from https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12528-Artificial-Intelligence-Act
- European Commission. (2022). Proposal for a Regulation of the European Parliament and of the Council on enhancing cooperation amongst member states in the field of artificial intelligence. Retrieved from https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12529-Enhancing-cooperation-amongst-Member-States-in-the-field-of-Artificial-Intelligence
- European Commission. (2022). Proposal for a Regulation of the European Parliament and of the Council on the establishment of a European Artificial Intelligence Board. Retrieved from https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12530-Establishment-of-a-European-Artificial-Intelligence-Board
- European Commission. (2022). Proposal for a Regulation of the European Parliament and of the Council on the establishment, operation and use of a European database on artificial intelligence. Retrieved from https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12531-Establishment-of-a-European-database-on-artificial-intelligence
- European Commission. (2022). Proposal for a Regulation of the European Parliament and of the Council on the European Union's list of AI-based products, services and acts subject to restriction or prohibition. Retrieved from https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12532-Proposal-for-a-Regulation-of-the-European-Parliament-and-of-the-Council-on-the-European-Union-s-list-of-AI-based-products-services-and-acts-subject-to-restriction-or-prohibition_en
- Reuters. (2022). Meta and OpenAI criticise EU's draft AI rules, threaten legal action. Retrieved from https://www.reuters.com/business/media-telecom/meta-openai-criticise-eus-draft-ai-rules-threaten-legal-action-2022-04-21/
- In line with the AI Act, technology companies like OpenAI and Anthropic, responsible for AI models such as ChatGPT and Claude, must adhere to strict transparency and copyright obligations when placing their models on the EU market.
- Under the AI Act, social networks like Facebook, Instagram, YouTube, and LinkedIn in the EU are required to clearly label AI-generated contents to prevent misunderstandings and maintain transparency.
