The US Artificial Intelligence Safety Institute has announced agreements with OpenAI and Anthropic to review their AI models for capabilities and safety. Under the terms of the memorandum of understanding with two prominent AI startups, the government agency will research, test, and evaluate their AI models.
This agreement underscores the varied strategies of government agencies in managing AI’s potential risks. While Europe focuses on implementing the AI Act, US regulators are taking a more collaborative approach by working closely with companies.
US AI Safety Institute to have early access to new AI models
The AI Safety Institute has disclosed that the deal allows it to have early access to all major new AI models from the two companies before their public release. This will enable the institute to evaluate the capabilities and risks of the AI models before their public availability.
However, the institute will maintain access even after the AI models are available to the public. The research and evaluation process will also include working with the companies on methods to address the identified risks.
Elizabeth Kelly, Director of the US AI Safety Institute, described the deal as the beginning of a long-term effort to develop a framework for responsible AI. She said:
“Safety is essential to fueling breakthrough technological innovation. With these agreements in place, we look forward to beginning our technical collaborations with Anthropic and OpenAI to advance the science of AI safety.”
The AI Safety Institute is part of the US Department of Commerce’s National Institute of Standards and Technology (NIST). Established through an executive order in 2023, the agency will also partner with the United Kingdom AI Safety Institute on the research and evaluation process.
Apple is considering investing in OpenAI
Meanwhile, Apple reportedly plans to invest in OpenAI as the startup prepares for its next funding round, which will value it at $100 billion. According to media reports, Thrive Capital will lead the fundraising with $1 billion, and OpenAI’s biggest investor, Microsoft, is also planning to participate in the round.
Microsoft’s interest is quite notable, given its 49% stake in OpenAI. However, Apple’s potential investment is less surprising, considering the tech giant has been working with the ChatGPT developer to integrate its technologies into the highly anticipated Apple Intelligence systems.
Given that AI might represent the next significant advancement for iPhones, Apple’s interest in securing a stake in one of the leading AI firms is logical. Apple rarely invests in startups, but when it does, it targets those with crucial technologies that could enhance its business.
However, this deal could limit Apple’s ability to integrate generative AI models from other companies into its systems. Apple has previously engaged with potential partners like Google Gemini, Anthropic, Perplexity, and Meta. Although it chose to begin with OpenAI due to its superior technology, a potential investment might restrict Apple from incorporating other models, posing a risk of over-reliance on a single source.
Source: https://www.cryptopolitan.com/openai-us-govt-partnership-apple-investment/