- Palo Alto Networks CEO Nikesh Arora says the industry is unlikely to coordinate a slowdown in AI development, arguing that companies should focus instead on building safer systems and stronger security protections.
Palo Alto Networks Inc. (NASDAQ: PANW) CEO Nikesh Arora said slowing the development of artificial intelligence is “unrealistic,” adding to a growing debate among technology executives over how quickly frontier AI should advance.
In an interview with CNBC’s The Tech Download podcast recorded Tuesday, Arora said companies are unlikely to move at the same pace because some developers will continue pushing frontier AI capabilities even if others voluntarily slow down.
Arora said the focus should instead be on whether AI products are safe enough to release and whether appropriate security measures and guardrails are in place.
His comments come as AI companies and technology executives take increasingly different positions on the risks associated with more powerful models. Anthropic CEO Dario Amodei has argued that the industry should slow the pace of AI development, while Nvidia CEO Jensen Huang has taken a more skeptical view of calls for a development slowdown.
Why Nikesh Arora Says an AI Slowdown Would Be Difficult
Arora's argument centers on coordination. If companies, countries or individual developers follow different approaches to AI development, he said, attempting to impose a common pace would be difficult.
He also rejected the idea that AI poses a 10% probability of causing human extinction, describing that possibility as extremely small while emphasizing the need for safety and security measures. That is Arora's assessment rather than an established scientific consensus.
The disagreement has broader implications for the AI industry. A coordinated slowdown could affect the pace of model development and the enormous investment going into computing infrastructure. Continued development, meanwhile, increases the importance of cybersecurity as companies deploy AI agents and increasingly autonomous software.
The issue is particularly relevant to Palo Alto Networks because the company sits on the security side of the AI economy.
Palo Alto Networks reported $3.41 billion in fiscal fourth-quarter revenue, up 34% year over year, while next-generation security annual recurring revenue increased 63% to $9.1 billion. The company expects fiscal 2027 revenue of $14.1 billion to $14.2 billion, representing 23% to 24% growth.
Palo Alto Networks Is Using AI to Defend Against AI-Powered Attacks
Just two days before Arora's CNBC interview, Palo Alto Networks introduced Unit 42 Continuous Frontier AI Defense, a subscription service designed to continuously identify, validate, and help remediate security vulnerabilities.
The service uses multiple AI models, including Anthropic's Claude Mythos 5 and OpenAI's GPT-5.6-Cyber, alongside open-weight models. Palo Alto says its system can test web applications, APIs, cloud infrastructure, source-code repositories and network assets as those environments change.
The company said its testing across more than 100 customer engagements found exposures at every customer assessed, with 37% rated high or critical. Palo Alto also said its internal testing identified a year's worth of exposures in three weeks. Those figures are company-reported results rather than independently audited industry benchmarks.
The timing highlights the commercial side of the AI-safety debate. Rather than assuming AI development will stop, cybersecurity companies are preparing for an environment in which increasingly capable AI systems are widely deployed, including by attackers.
Palo Alto said threat actors are already using AI to automate parts of cyberattacks, potentially compressing attack cycles from weeks to hours. Its new service is designed around that faster threat environment.