California Gov. signed AB 1709, a package of 13 laws to shield minors from harmful tech, including banning behaviorally addictive features for under-16s, mandating parental controls on AI chatbots, and a four-year ban on AI companion toys, as part of a broader online-safety push.

Industry feedback centers on Meta’s concerns about reduced personalization, though proponents argue safeguards enable age-appropriate content with guardrails.

The package also imposes penalties up to $1 million per child for negligence by large platforms that harm minors.

An accompanying report questions verification feasibility and cost, but provides a concrete example tied to a recent probe.

OpenAI and Anthropic endorse safeguards and third-party verification, distinguishing audits from evaluations in the safety framework.

The policy is framed as ongoing governance work by policymakers and industry, with historical context from thinkers like Friedman and Klein to shape the regulatory environment.

A central question is how rigorously the Commission writes standards; effectiveness hinges on the quality of criteria and enforcement rather than immediate model limits.

Experts warn the field relies on unproven tools and could overwhelm human oversight, underscoring safety concerns.

Real impact will depend on enforcement resources and whether measures translate into meaningful change.

OpenAI is weighing slower development due to existential-risk concerns, with calls for voluntary slowdowns and transparency; U.S. leadership priorities contrast with other global approaches.

OpenAI spends about a fifth of compute on safety monitoring, highlighting the ongoing cost of verification and oversight.

An e-Safety Advisory Commission is created to study recommendations, though critics say there’s insufficient representation from First Amendment experts.