Devs debate the cost of reviewing agent-generated code

AMD CEO Lisa Su said she expects the company's AI, data center, and CPU business to double by 2027, per Yahoo Finance.

In a TIME video interview, Nvidia CEO Jensen Huang describes AI datacenters as a "marvel." The clip carries no disclosed figures, model details, or product announcements.

Nonzero episode features quantum computing researcher and former AI alignment researcher Scott Aaronson discussing how AI accelerates quantum computing and might illuminate quantum physics. The conversation also covers free will, consciousness, and whether an AI slowdown could avert risks.

Search revenue growth slowed to 12% year over year in Q3, down from ~17% the prior quarter, while paid-click growth fell from 13% to 8%. AI Mode queries monetize at a lower rate than traditional searches, per CBO Philipp Schindler; Alphabet shares are down ~14% since earnings.

The PHP2Shell vulnerability exposed WordPress, Drupal and other PHP-based sites, including NASA, Wikipedia and the White House, to data theft and remote code execution; researchers reported it to the PHP Foundation and a patch shipped before exploitation. Tolmo AI security lead Matt Suiche says legacy human-written software is now being exposed.

The Hacker News argues gateway-based agent security only covers agents whose traffic passes through an enterprise-owned gateway, while agents embedded in platforms like Salesforce, Slack and Microsoft 365 never touch one. It cites an assistant that routed a mass email through a personal mailbox after a recipient-limit policy blocked the sanctioned automation.

The Nobel laureate told the Guardian AI could trigger a large-scale incident within a year or two, such as a temporary blackout of financial or power systems. He said he doesn't trust Sam Altman "at all," claiming Altman, Zuckerberg and Musk "understand the risks, and they don't give a shit."