California Governor Gavin Newsom issued an executive order to form an expert panel that will strengthen AI safety laws, push for third-party safety evaluations, require risk reporting, and explore an emergency shut-off for frontier AI models as part of a broader safeguard strategy.

The move places California among growing state-level AI regulatory efforts and aligns with discussions of a potential US-led global regulatory body for AI.

The policy debate spans the 2028 presidential landscape, with candidates and lawmakers weighing the need for robust safeguards amid public concern.

Implementation challenges cited include high testing costs and talent shortages, though these could be mitigated through collaborations and incremental technology upgrades.

Industry voices stress responsibility: leaders warn that neglecting alignment risks falling behind, while others stress safety and monitoring over feature speed amid competition.

Experts caution that kill switches alone may be outdated or easily gamed, arguing for broader third-party assessments, safety standards, and incentives to implement safeguards.

Some experts view a universal kill switch as aspirational or insufficient since AI operates across multiple servers and environments, complicating full control.

Congress debates the kill switch idea, with concerns about stifling innovation if enacted too soon.

Regulators emphasize practical tools like risk assessment, bias detection, data privacy, and model transparency to help firms meet standards and avoid liability.

Security vulnerabilities were highlighted when Hacktron researchers exploited a flaw in an OpenAI public forum, underscoring risks to AI-enabled services.

Europe faces tensions between digital sovereignty and competitiveness, seeking a balance between regulation and maintaining AI advancement.

Academic cautions against doomsday scenarios while urging regulation and cybersecurity improvements to prevent breaches.