AI Safety Slowdown Calls Shift Venture Capital Toward Security Startups

Anthropic and OpenAI's recent calls to deliberately slow frontier model development are shifting investor attention toward AI safety and infrastructure startups, according to venture capitalists tracking the sector. On September 12, 2026, Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier," proposing a three-part plan to throttle AI capability gains by one to two years to give safety research time to catch up. The proposal includes embedding permanent third-party evaluators from organizations like METR inside frontier labs with employee-level access, establishing industry coordination on safety standards among democratic nations, and pursuing bilateral coordination with authoritarian governments on narrowly defined dangerous uses.