OpenAI will let US government test safety of forthcoming model
What's the story
OpenAI, a leading artificial intelligence (AI) research organization, has committed to providing the US AI Safety Institute with early access to its forthcoming generative AI model. The announcement was made by OpenAI's CEO Sam Altman in a post on X. This move mirrors a similar agreement that OpenAI entered into with the UK's AI safety body in June. The US AI Safety Institute is a federal agency focused on evaluating and mitigating risks associated with AI platforms.
Twitter Post
Take a look at Altman's post
a few quick updates about safety at openai:
— Sam Altman (@sama) August 1, 2024
as we said last july, we’re committed to allocating at least 20% of the computing resources to safety efforts across the entire company.
our team has been working with the US AI Safety Institute on an agreement where we would provide…
Safety measures
A response to safety concerns
OpenAI's collaboration with US AI Safety Institute comes amid criticisms that the research lab has been prioritizing the development of more advanced generative AI technologies over safety measures. In May, OpenAI dissolved a team dedicated to creating safeguards against potentially harmful "superintelligent" AI systems, leading to the departure of senior executives Jan Leike and Ilya Sutskever. Sutskever was a co-founder of OpenAI and its chief scientist for nearly a decade.
Safety commitment
OpenAI's response to safety research criticism
In response to the criticism, OpenAI has announced plans to remove non-disparagement clauses that could potentially deter whistleblowing, and establish a safety commission. The organization also committed 20% of its computational resources toward safety research. However, these measures were met with skepticism after it was revealed that the safety commission was staffed by company insiders and a senior AI safety executive was reassigned.
Policy endorsement
OpenAI's commitment amid legislative scrutiny
The commitment to the US AI Safety Institute coincides with a period when five senators have questioned OpenAI's policies in a letter to Altman. In response, Jason Kwon, OpenAI's Chief Strategy Officer, stated that the organization is "dedicated to implementing rigorous safety protocols at every stage of our process." This commitment also aligns with OpenAI's support for the Future of Innovation Act, a proposed Senate bill that would empower the Safety Institute to set standards and guidelines for AI models.
Policy impact
OpenAI's influence on federal AI policy
Altman serves on the US Department of Homeland Security's Artificial Intelligence Safety and Security Board, which advises on safe and secure AI development and deployment. OpenAI has also significantly increased its federal lobbying expenditures this year, spending $800,000 in the first half of 2024 compared to $260,000 in all of 2023. These developments have led to speculation about OpenAI's potential influence over federal AI policy-making.