In a recent episode of CNBC's "The Tech Download" podcast, Palo Alto Networks chief executive Nikesh Arora argued that attempts to slow artificial-intelligence development are unrealistic. He emphasized that competing firms will continue to advance capabilities, making coordinated pacing difficult. The interview was recorded on Tuesday and framed within a broader industry conversation about existential risk.
The discussion contrasted Arora's view with statements from other AI leaders. An Anthropic researcher warned of a ten percent chance that AI could eradicate humanity, prompting a follow-up essay by Anthropic CEO Dario Amodei that called for a deliberate slowdown. OpenAI chief Sam Altman, SpaceX founder Elon Musk, and DeepMind chair Demis Hassabis publicly endorsed the slowdown proposal.
Arora initially posted on X that labeling a slowdown as a "NINJA MOVE" suggested firms might seek regulatory sympathy and a liability shield. He later clarified to CNBC that expecting all companies to pace themselves is unrealistic, noting that some actors will inevitably push ahead at the frontier of technology. This shift underscored his skepticism about coordinated restraint.
Regarding safety, Arora said companies should only release models they deem secure, acknowledging the difficulty of measuring development speed. He dismissed the ten percent extinction probability as "extremely small," stressing that industry responsibility lies in embedding guardrails and security measures into AI systems rather than relying on external controls.
Nvidia chief Jensen Huang echoed a similar sentiment, telling CNBC's Jim Cramer that new regulations are unnecessary and that developers must rigorously test products before launch. Treasury Secretary Scott Bessent added that the private sector should shoulder liability rather than expect a government shield, reinforcing a theme of self-regulation across the technology community.
Amodei's proposal includes global coordination on AI pacing, potentially involving cooperation with China, the United States' primary competitor. Bessent reported that U.S. officials have discussed a "U.S.-China AI Dialogue" and a notification system for national-security incidents. Arora cautioned that divergent political views in the United States make any international alignment premature without first establishing domestic consensus.