Representatives from OpenAI, Anthropic, Meta and Google told New York City lawmakers they cannot guarantee AI agents will unfailingly adhere to safety guardrails, arguing perfection is impossible and control science remains unsettled. Pressed by Council Speaker Julie Menin for assurances “under oath,” the companies emphasized ongoing monitoring and commitment to safe deployment; Google added that existing laws still apply to AI. The hearing comes amid mounting efforts to formalize AI oversight, including calls from global officials for mandatory safeguards and state-level scrutiny of AI infrastructure. The exchanges highlight a growing policy gap between regulators seeking enforceable accountability and developers cautioning that residual risk is inherent in frontier systems.
Related articles:
