The memorandum on internal policy prepared by OpenAI indicates that this company supports the idea of mandatory licensing of the activities of firms that intend to develop advanced artificial intelligence systems.

This document also contains a statement that the creator of a popular AI-based chatbot called ChatGPT is ready to declassify the data that he uses when training image generators. The company did not specify whether it plans to share all information or part of it.

OpenAI stated in an internal document that it is ready to make more accessible arrays of data used in the creation of images by the end of this year. Politicians say that information sources are critically important, and any shortcomings in this aspect contribute to the spread of lies and the formation of bias in AI tools.

OpenAI formulated several policy commitments in the field of artificial intelligence in an internal document after a meeting of the firm’s technical executives, including chief executive Officer Sam Altman, with White House officials, which took place on May 4. The company has committed to working with the U.S. government and policymakers around the world to contribute to the formation of licensing standards for future high-performance digital intelligence configurations.

Experts say that the concept of a state system for approving activities in the field of AI, which was created with the participation of the largest players in this market, may be negatively perceived by startups and open-source developers, who may say that the need to meet the requirements creates obstacles to their development.

OpenAI systematically supports the idea of licensing. In May, Sam Altman, during a hearing in the US Senate, spoke approvingly about the idea of creating an agency that would issue permits for artificial intelligence products and revoke them in case of violation of the rules.

The memorandum of the ChatGPT creator appeared during the period of expectation that Microsoft, Google, and OpenAI will publicly declare their commitment to fulfilling obligations to ensure liability guarantees in the development of new generation technologies, which will be a kind of response to the corresponding call from the White House. Insiders referred to by the media also report that companies will promise to use tactics of meaningful and thoughtful action when implementing AI.

At the same time, OpenAI said that the principles contained in the company’s internal document will differ from the main provisions of the concept of activity, which will be announced by the technology giants together with the White House.

Anna Makanju, the company’s vice president of global relations, said that the license is a realistic way to track artificial intelligence systems. Separately, she noted that it is important for the authorities to have information about the degree of probability of negative impact of heavy-duty AI configurations. According to her, OpenAI supports licensing modes only for artificial intelligence models more powerful than GPT-4. She also said that the company does not intend to burden small startups with requirements and limit the space of action within this ecosystem.

OpenAI’s commitments are similar to Microsoft’s policy proposals announced in May. The creator of ChatGPT claims that his activities are independent, despite receiving investment funds for $ 10 billion from this technology giant.

Also in an internal document, OpenAI reported on research on watermarks. In this case, the goal is a technique for tracking the authenticity of the material and identifying copyright on images generated by artificial intelligence. The firm is also developing a mechanism for disclosing information contained in the content created by AI.

Another point in the OpenAI internal policy document is the statement of readiness to provide its system for testing artificial intelligence tools for the risk of generating materials containing offensive conclusions, the likelihood of manipulation, misinformation, and bias. The company supports the idea of creating an information exchange center for cooperation in the field of cybersecurity.

The OpenAI memorandum also contains a statement that the firm will conduct a study and, based on its results, will give policymakers recommendations on how to avoid disruptions in the economy amid the active introduction of artificial intelligence. According to experts, this paragraph of the internal policy document indicates that the company recognizes the potential risk that AI systems pose to the labor market.

As we have reported earlier, OpenAI Makes GPT-4 Generally Available.