Watchtower tracks changes to corporate and government AI safety policies, both announced and unannounced. Click any entry for details.
Jul 2, 2024
Magic.dev
Moderate
On July 2, 2024, Magic.dev released the first version of their AGI Readiness Policy, a framework that has similarities to the Responsible Scaling Policies released by Anthropic, OpenAI, and Google.
In it, they announce plans to monitor for risks across four categories, and to implement a full, detailed responsible scaling policy upon reaching a threshold in capabilities as measured by LiveCodeBench or a set of private benchmarks.