Editorial companion

What stands out

Executives from Google, OpenAI, Anthropic and Meta testified live before the New York City Council about AI risks, focusing on public safety, misinformation and where responsibility sits as models scale. The most striking element is the public, adversarial framing, lawmakers pressed companies on concrete safeguards and real-world harms rather than abstract promises, and the companies offered a mix of policy proposals and technical descriptions. That back-and-forth highlights regulators moving from questions to demands, and it forced company spokespeople to translate complex model behavior into specific mitigation steps on the record.