Nvidia says its AI chips don't have a 'kill switch' after Chinese accusation

Nvidia says its AI chips don't have a 'kill switch' after Chinese accusation

On Tuesday, Nvidia firmly dismissed allegations from China claiming that its data center GPUs for artificial intelligence are equipped with a remote deactivation feature, often referred to as a 'kill switch.' In a blog post, Chief Security Officer David Reber emphasized, "NVIDIA GPUs do not and should not have kill switches and backdoors." This statement follows the Cyberspace Administration of China's recent request for Nvidia to submit documentation regarding what it described as security vulnerabilities in the H20, Nvidia's AI chip tailored for the Chinese market. The Chinese regulator specifically pointed to potential 'backdoor' security risks, according to reports. Nvidia's response highlights the complexities the company faces amid geopolitical tensions, as its AI chips continue to be sought after by nations and businesses globally. In the U.S., lawmakers have suggested new regulations that would require AI chips designated for export to include location-tracking capabilities. The U.S. has already imposed export controls on certain Nvidia chips to China, citing national security concerns, including fears that these chips could be utilized to enhance China's military or AI capabilities. Nvidia CEO Jensen Huang has previously argued that it would be beneficial for the U.S. if Nvidia's chips became the global standard for AI computing, particularly among developers in China. The H20 chip generates significant revenue for Nvidia, reportedly contributing billions in sales each quarter, although the company does not typically disclose specific revenue figures for individual products. In April, exports of the H20 to China were temporarily halted, and Nvidia indicated that its guidance would have been approximately $8 billion higher were it not for these restrictions on sales. In July, the Trump administration announced it would issue a waiver allowing the resumption of sales for these chips. Experts in Silicon Valley, including security professionals, generally criticize the concept of backdoors in technology, as they pose inherent risks. Apple has similarly resisted government pressures to create what it refers to as 'backdoors' in its products. In his blog post, Reber highlighted that hidden backdoors are not just a risk for government misuse but could also be exploited by cybercriminals, contradicting the core tenets of cybersecurity. He warned that incorporating a kill switch or backdoor into products like Nvidia's GPUs would endanger U.S. national security, stating, "Hardwiring a kill switch into a chip is something entirely different: a permanent flaw beyond user control, and an open invitation for disaster." Reber's analogy compared this risk to purchasing a car while the dealership retains a remote control for the parking brake, indicating a severe loss of control for the consumer.

Sources : CNBC

Published On : Aug 05, 2025, 18:20

Cybersecurity
Privacy Concerns Emerge as Meta Workers View Sensitive Footage from Ray-Ban Smart Glasses

Meta's commitment to user privacy is facing intense criticism after a recent report revealed that employees at a subcont...

Ars Technica | Mar 05, 2026, 23:40
Privacy Concerns Emerge as Meta Workers View Sensitive Footage from Ray-Ban Smart Glasses
AI
Anthropic's CEO Engages Pentagon Amid Supply Chain Controversy

Dario Amodei, the CEO of Anthropic, is not stepping back from discussions with the Pentagon, despite the company facing ...

Business Insider | Mar 06, 2026, 02:30
Anthropic's CEO Engages Pentagon Amid Supply Chain Controversy
Computing
Navigating the AI Landscape: Insights from Morgan Stanley's Tech Conference

The recent Morgan Stanley Tech, Media, and Telecom conference showcased a formidable lineup of industry leaders, includi...

CNBC | Mar 05, 2026, 23:25
Navigating the AI Landscape: Insights from Morgan Stanley's Tech Conference
AI
Microsoft to Maintain Access to Anthropic's AI Solutions Amid Security Concerns

On Thursday, Microsoft announced it will continue to offer Anthropic's artificial intelligence technologies to its clien...

CNBC | Mar 06, 2026, 01:15
Microsoft to Maintain Access to Anthropic's AI Solutions Amid Security Concerns
Cybersecurity
Meta's AI Smart Glasses Under Fire: Lawsuit and Regulatory Scrutiny Over Privacy Violations

Meta's AI-enhanced Ray-Ban smart glasses are currently at the center of a legal storm due to serious privacy concerns. R...

Business Today | Mar 06, 2026, 06:05
Meta's AI Smart Glasses Under Fire: Lawsuit and Regulatory Scrutiny Over Privacy Violations
View All News