Breaking

Nvidia CEO Jensen Huang Says AI Safety Needs Engineering, Not New Laws

September 19, 2026 at 08:02 EDT

AI Agents
Policy & Regulation
Infra & Chips

Nvidia CEO Jensen Huang has rejected calls to slow frontier AI development, saying safety should be solved through engineering discipline rather than new laws or industry-wide speed limits. The stance, made at Salesforce Dreamforce and echoed in an interview scheduled for broadcast this weekend, puts the AI chipmaker firmly on one side of a fast-widening governance fight.

Sep. 2026 · Nvidia

AI safety: build fast, verify faster

Jensen Huang argues that engineering discipline, release readiness and market pressure can handle AI safety without new slowdown laws. Rivals and safety advocates want independent checks before agents become harder to control.

>$5T

Nvidia market cap context

26%

Claude-led model R&D share

100x+

Drop cited with stronger harnessing

Steps in pacing proposal

What stronger engineering controls are meant to change

A production harness and system prompt were cited as reducing infrastructure-compromise propensity by more than 100x.

Huang's operating logic

Build fast

Keep innovation moving.

→

Test deeply

Use sandboxes and monitoring.

→

Pause unsafe

Hold releases that are not ready.

→

Ship only ready

Make safety a release gate.

Engineering case

Builders own release risk

Supporters argue that teams closest to the systems can test, monitor, pause and fix faster than a new legal regime can move.

Governance case

Verification is the gap

Critics want external evaluators and shared standards because agent failures can affect outsiders before a company catches them.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.