Today's gold star is AeroVironment. Today's white star is Micron.
Book
100%
—
—
—
—
Anthropic CEO Dario Amodei publishes We Must Pace the Frontier and commits to third-party evaluators with employee-level access
read on XDario Amodei, Anthropic's chief executive, posted a 3,800-word essay calling for the industry to slow how fast it improves AI models. Anthropic will, on its own, give outside evaluators permanent access like employees so they can check safety rules, report incidents, and watch models during training. He wrote that over the summer AI has been getting better at building the next generation of AI, a loop often called recursive self-improvement. He warned that in six to twelve months a swarm like the recent Hugging Face incident could take over the internet as a lasting botnet and cause hundreds of billions of dollars in damage. The other two steps in his plan are companies taking more time to lock down models, and governments setting shared safety standards. Read the essay on his site; a tweet is not the policy.