Last year, as part of a comprehensive review of NIH’s intramural policies, I announced that NIH was adopting an updated agency-wide framework to embed principles of academic freedom within our Intramural Research Program. This framework recognizes that intramural researchers must be free to share their scientific viewpoints and to communicate their scientific findings through manuscripts, presentations, media interviews, and so on. This includes hosting outside lecturers for events focused on their scientific work. Today, NIH reaffirms this commitment to academic freedom as members of the intramural research community host Kristian G. Andersen, MPhil, PhD, as a guest lecturer. Dr. Andersen is a co-author of a March 2020 paper published in Nature Medicine titled “The proximal origin of SARS-CoV-2.” The paper’s authors assert: “Our analyses clearly show that SARS-CoV-2 is not a laboratory construct or a purposefully manipulated virus.” They further write that “we do not believe that any type of laboratory-based scenario is plausible.” The paper was cited extensively by members of the media, social media companies, and government officials to refute or suppress lab-origin hypotheses and to discredit anyone who publicly dissented from the dominant natural-origins narrative. As then-NIH Director Dr. Francis Collins wrote, “this study leaves little room to refute a natural origin for COVID-19.” Yet before and after the paper’s publication, Dr. Andersen himself expressed substantially more uncertainty. In January, he wrote in an email that some features of the virus “(potentially) look engineered.” One month after the paper’s publication, he wrote privately that he was not fully convinced that every lab-based scenario could be ruled out. In January 2025, the Central Intelligence Agency released the results of an assessment conducted during the Biden administration that found a “research-related origin” of SARS-CoV-2 was “more likely than a natural origin,” while assessing both theories as plausible. Given the evidence I’ve seen, I think the preponderance of evidence supports the theory that the virus leaked from a lab. I am keenly aware that no hypothesis about the origins of SARS-CoV-2 has been conclusively proven. When it comes to open scientific questions like this, our ability to pursue truth depends on our freedom to explore various hypotheses, weigh all available evidence, and deliberate openly. This freedom was denied to many scientists during the pandemic. 𝗜 𝗮𝗺 𝗽𝗹𝗲𝗮𝘀𝗲𝗱 𝘁𝗵𝗮𝘁 𝗮𝗹𝗹 𝗡𝗜𝗛 𝘀𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁𝘀 𝗮𝗻𝗱 𝘁𝗵𝗲𝗶𝗿 𝗴𝘂𝗲𝘀𝘁𝘀—𝗻𝗼 𝗺𝗮𝘁𝘁𝗲𝗿 𝘄𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝘁𝗵𝗶𝗻𝗸 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗼𝗿𝗶𝗴𝗶𝗻𝘀 𝗼𝗳 𝗦𝗔𝗥𝗦-𝗖𝗼𝗩-𝟮—𝗲𝗻𝗷𝗼𝘆 𝗶𝘁 𝗻𝗼𝘄.

Mustafa Suleyman (head of AI at Microsoft, DeepMind co-founder) on Anthropic training Claude to wonder if it's conscious: "They have expressed uncertainty about the basic nature of Claude as a new kind of entity… but they think that it is a significant enough possibility that in their training document, they've repeatedly said they want to try to improve the well-being of Claude… Anthropic, has published a constitution... This document is written to Claude and is seen by Claude and used to train Claude. The problem I have with this is that this speculation has been baked into the very training of Claude, and therefore Claude can only reproduce that ambiguity when you talk to it. So today, Claude is speaking to tens or hundreds of millions of people every week, and some of those people are asking whether or not Claude is conscious. And it is saying, 'Well, I'm not sure,' precisely because that's what's been trained into it."

@bgurley @elizabethjdias @nytimes Seems like Anthropic’s approach is vastly different than the other frontier labs, where the machine will inevitably control humans and not the other way around. Not a good approach

“One of the main lessons to come out of Tetlock’s work is that the best generalist forecasters are usually better at telling the future than domain-specific experts, even when it comes to those experts’ areas of study. This is because forecasting itself is its own science that requires a unique way of thinking, and a person having a great deal of knowledge about a field is no guarantee that they have good epistemological habits. If you’re going to ask, for example, whether American democracy is falling apart, an expert might start by making a superficial comparison between the modern US and the Weimar Republic. Someone with a forecasting mindset would be more likely to begin with a question like “how often do advanced democracies slide into dictatorships?” The answer is not very often.” Important reminder from @RichardHanania that domain experts are not the best at predicting what happens, even in their own domains.

Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry. Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come. But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility. This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems. Together, we are building the foundation of the AI economy. Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world.