Skip to main content
Add as a preferred source on Google


NIST's U.S. AI Safety Institute to Collaborate with Anthropic and OpenAI on AI Research

  • By John K. Waters
  • 09/05/24

The U.S. AI Safety Institute, part of the National Institute of Standards and Technology (NIST), is partnering with AI companies Anthropic and OpenAI to collaborate on AI safety research, testing, and evaluation.

Under the Memoranda of Understanding, the U.S. AI Safety Institute will gain access to new AI models from both companies before and after their public release. This collaboration aims to assess the capabilities and risks of these models and develop methods to mitigate potential safety concerns.

"Safety is essential to fueling breakthrough technological innovation," said Elizabeth Kelly, director of the U.S. AI Safety Institute, in a statement. "With these agreements in place, we look forward to beginning our technical collaborations with Anthropic and OpenAI to advance the science of AI safety."

"These agreements are just the start," she added, "but they are an important milestone as we work to help responsibly steward the future of AI."

The U.S. AI Safety Institute also intends to work closely with its partners at the U.K. AI Safety Institute to offer feedback to Anthropic and OpenAI on potential safety enhancements to their models.

"Safe, trustworthy AI is crucial for the technology's positive impact," said Anthropic co-founder and head of policy Jack Clark, in a statement. "Our collaboration with the U.S. AI Safety Institute leverages their wide expertise to rigorously test our models before widespread deployment."

The agreements come at a time of increasing regulatory scrutiny over the safe and ethical use of AI technologies. California legislators are also poised to vote on a bill regulating AI development and deployment.

The initiative builds on NIST’s longstanding legacy in advancing measurement science and standards, with the aim of fostering the safe, secure, and trustworthy development and use of AI, as outlined in the Biden-Harris administration’s Executive Order on AI.

"We believe the institute has a critical role to play in defining U.S. leadership in responsibly developing artificial intelligence," said OpenAI chief strategy officer Jason Kwon, "and hope that our work together offers a framework that the rest of the world can build on."

About the Author

John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS.  He can be reached at [email protected].

Featured

  • lock icons, shields, and glowing data paths

    Studies Suggest AI Is Accelerating Familiar Cyber Attacks

    Research released around the recent Black Hat USA 2026 conference demonstrates that artificial intelligence is increasing the speed and scale of cybersecurity activity without replacing the attack methods defenders already face.

  • Digital Ai padlock locked for computing system on dark green background

    Microsoft, Nvidia Push Enterprise AI into Active Cyber Defense

    AI security is entering a new stage as enterprise organizations look beyond protecting models from prompt injection, unauthorized access, and data exposure.

  • Neon blue security locks with a single red highlight

    With AI, Cybersecurity Focus Shifts from Finding Flaws to Fixing Them

    For decades, one of cybersecurity's biggest challenges has been finding vulnerabilities before attackers do. A growing number of security professionals now say artificial intelligence is changing that equation, shifting the focus from discovering flaws to fixing them quickly enough to prevent exploitation.

  • circuit patterns

    Anthropic Intros Lower-Cost Claude Sonnet 5

    Anthropic has launched Claude Sonnet 5, positioning the model as its most autonomous mid-tier offering to date and a lower-cost alternative to its flagship Opus 4.8 system. The company said the model can plan multi-step tasks, operate tools such as browsers and terminals, and complete agentic work at a level that previously required larger and more expensive models.