Compliance as a Barrier: California Introduces Safety Filter for AI State Contracts

The free market of AI solutions is facing harsh bureaucratic reality. On March 31, 2026, the Governor of California signed an executive order that radically changes the rules of the game for tech giants. From now on, any company seeking state contracts must prove the existence of built-in safeguards against abuse and algorithmic bias (harmful bias).

This order is the first precedent for the large-scale legalization of AI Safety principles at the level of multi-billion dollar state budgets. For vendors like OpenAI or Anthropic, this means that "black boxes" will no longer pass the audit. As we predicted (in the context of the Gartner report on the growth of Explainable AI), transparency and observability tools are becoming the ticket to government money. Companies unable to mathematically prove the safety of their models will simply be left out of California's public procurement.

Source: Governor of California / Reuters