OpenAI has signed the first-of-its-kind agreement with the U.S. government to collaborate on AI safety research and assessment.
The deal comes as the corporate is reportedly in talks to boost funding valued at a staggering $100 billion, underscoring the staggering pace of development and investment within the AI sector.
On Thursday, the United States Institute for Security within the Area of Artificial IntelligencePart of the National Institute of Standards and Technology (NIST), announced that it had reached agreements with each OpenAI and rival AI company Anthropic.
The partnerships will give the federal government insider access to vital latest AI models before and after they’re released, although exactly what meaning is difficult to discern. It's fair to say that even the AI developers themselves don't really understand how their models work.
We are pleased in regards to the agreement with the US AI Safety Institute for pre-testing of our future models.
For many reasons, we imagine it will be important that this happens at a national level. US must proceed to steer!
— Sam Altman (@The same) 29 August 2024
“Safe, trustworthy AI is critical to the technology's positive impact,” said Jason Kwon, chief strategy officer at OpenAI. “We imagine the institute plays a critical role in defining U.S. leadership within the responsible development of artificial intelligence.”
Elizabeth Kelly, director of the US AI Safety Institute, called the agreements “a very important milestone in our efforts to responsibly shape the longer term of AI.”
As a part of the agreement, the institute will provide OpenAI with feedback on possible safety improvements to its models, working closely with the UK-based AI Safety Institute, which also Access requested how AI models work.
The agreements have been a protracted time coming as a part of the Biden administration's efforts to control AI.after Biden's Implementing Regulationsigned in October 2023, AI governance within the U.S. has slowly been gaining momentum, although some would argue that progress leaves much to be desired.
The timing of the agreement is notable for 2 reasons. First, it comes at a time when OpenAI is reportedly in talks to start a brand new round in financing that may increase the corporate's value to over $100 billion.
This astronomical figure represents greater than a threefold increase from the $29 billion valuation reported last yr.
According to sources accustomed to the matter, enterprise capital firm Thrive Capital is predicted to steer the round with an investment of $1 billion. Tech giant Microsoft, already OpenAI's biggest backer, can also be expected to take part in the financing.
Second, OpenAI is reportedly about to release a brand new product, GPT -5, SearchGPTor one other iteration codenamed “Project Strawberry.”
Project Strawberry, originally called Q*, reportedly combines an AI model with an autonomous AI agent that may surf the Internet.
Crucially, OpenAI reportedly demonstrated Strawberry to US security officials, which can have been a part of this latest take care of the AI Safety Institute and NIST.
All in all, things have been pretty quiet at OpenAI. GPT-4o was touted as a significant advance, but its crown jewel – the voice chat feature – has not yet been rolled out on a big scale. OpenAI cited security and regulatory hurdles as the rationale for the delay.
Could OpenAI enter into this latest partnership to avoid such problems in the longer term?
