Structured capability access ("SCA") is an emerging paradigm for the safe deployment of artificial intelligence (AI). Instead of openly disseminating AI systems, developers facilitate controlled, arm's length interactions with their AI systems. The aim is to prevent dangerous AI capabilities from being widely accessible, whilst preserving access to AI capabilities that can be used safely. The developer must both restrict how the AI system can be used, and prevent the user from circumventing these restrictions through modification or reverse engineering of the AI system. SCA is most effective when implemented through cloud-based AI services, rather than disseminating AI software that runs locally on users' hardware. Cloud-based interfaces provide the AI developer greater scope for controlling how the AI system is used, and for protecting against unauthorized modifications to the system's design. This chapter expands the discussion of "publication norms" in the AI community, which to date has focused on the question of how the informational content of AI research projects should be disseminated (e.g., code and models). Although this is an important question, there are limits to what can be achieved through the control of information flows. SCA views AI software not only as information that can be shared but also as a tool with which users can have arm's length interactions. There are early examples of SCA being practiced by AI developers, but there is much room for further development, both in the functionality of cloud-based interfaces and in the wider institutional framework.

The history of science and technology shows that seemingly innocuous developments in scientific theories and research have enabled real-world applications with significant negative consequences for humanity. In order to ensure that the science and technology of AI is developed in a humane manner, we must develop research publication norms that are informed by our growing understanding of AI's potential threats and use cases. Unfortunately, it's difficult to create a set of publication norms for responsible AI because the field of AI is currently fragmented in terms of how this technology is researched, developed, funded, etc. To examine this challenge and find solutions, the Montreal AI Ethics Institute (MAIEI) collaborated with the Partnership on AI in May 2020 to host two public consultation meetups. These meetups examined potential publication norms for responsible AI, with the goal of creating a clear set of recommendations and ways forward for publishers. In its submission, MAIEI provides six initial recommendations, these include: 1) create tools to navigate publication decisions, 2) offer a page number extension, 3) develop a network of peers, 4) require broad impact statements, 5) require the publication of expected results, and 6) revamp the peer-review process. After considering potential concerns regarding these recommendations, including constraining innovation and creating a "black market" for AI research, MAIEI outlines three ways forward for publishers, these include: 1) state clearly and consistently the need for established norms, 2) coordinate and build trust as a community, and 3) change the approach.

After nearly a year of suspense and controversy, any day now the team of artificial intelligence (AI) researchers at OpenAI will release the full and final version of GPT-2, a language model that can "generate coherent paragraphs and perform rudimentary reading comprehension, machine translation, question answering, and summarization--all without task-specific training." When OpenAI first unveiled the program in February, it was capable of impressive feats: Given a two-sentence prompt about unicorns living in the Andes Mountains, for example, the program produced a coherent nine-paragraph news article. At the time, the technical achievement was newsworthy--but it was how OpenAI chose to release the new technology that really caused a firestorm. There is a prevailing norm of openness in the machine learning research community, consciously created by early giants in the field: Advances are expected to be shared, so that they can be evaluated and so that the entire field advances. However, in February, OpenAI opted for a more limited release due to concerns that the program could be used to generate misleading news articles; impersonate people online; or automate the production of abusive, fake or spam content.

We're releasing the 774 million parameter GPT-2 language model after the release of our small 124M model in February, staged release of our medium 355M model in May, and subsequent research with partners and the AI community into the model's potential for misuse and societal benefit. We're also releasing an open-source legal agreement to make it easier for organizations to initiate model-sharing partnerships with each other, and are publishing a technical report about our experience in coordinating with the wider AI research community on publication norms. To date, there hasn't been a public release of a 1558M parameter language model, though multiple organizations have developed the systems to train them, or have publicly discussed how to train larger models. For example, teams from both NLP developer Hugging Face and the Allen Institute for Artificial Intelligence (AI2) with the University of Washington have explicitly adopted similar staged release approaches to us. Since February, we've spoken with more than five groups who have replicated GPT-2[1].