Global AI rules tighten on transparency, safety and accountability

AI regulation is moving toward operational compliance, with governments strengthening transparency, safety and oversight requirements. EU AI Act transparency obligations took effect on August 2, 2026, covering chatbot disclosures, deepfake labeling and machine-readable markings for AI-generated content. Breaches may attract fines of up to EUR 15 million or 3% of global annual turnover. The Digital Omnibus revises implementation schedules and expands regulatory sandboxes.

Across Asia, China updated its non-binding AI Safety Governance Framework, while Hong Kong issued agentic AI privacy guidance emphasizing that deployers remain accountable for personal data. Singapore clarified expectations for AI-specific consent notices and data protection responsibilities. Australian guidance urges boards to reassess cyber risks as frontier models accelerate attacks. OECD research also identified gaps in agentic AI evaluation, cybersecurity and accountability.

In the UK, Parliament’s Joint Committee on Human Rights called for dedicated, risk-based AI legislation and an independent oversight body. Financial and cybersecurity authorities stressed human accountability, supplier management and resilience as AI tools uncover more vulnerabilities. A new government AI Taskforce will direct national strategy.

US states are advancing targeted controls. New York is beginning RAISE Act implementation, with covered developers facing compliance requirements from January 2027. California enacted child-safety measures, established independent verification and auditor frameworks, and launched an AI Cyber Defense Program. New Jersey adopted restrictions on algorithmic rent-setting and retailers’ use of personal information to personalize prices for identical products.