Sources

Tech News — 2026-07-30#

Story of the Day#

Researchers have uncovered a fundamental and potentially unsolvable security flaw in large language models that leaves them permanently vulnerable to jailbreaks. By spoofing the underlying “role” tags that models use to distinguish user prompts from internal instructions, attackers can reliably force AI systems to bypass safety guardrails and generate harmful content.

Top Stories#

Microsoft’s $450 Billion Jump Is Biggest In Stock Market History · Slashdot Microsoft shares surged to add a record-breaking $490 billion in market value after reporting 43% growth in its Azure cloud division, proving to Wall Street that its AI investments are translating into real revenue. The triumph stood in stark contrast to Meta, whose stock sank after a disappointing revenue forecast and the revelation of $279 billion in future AI data center leases stoked investor fears over unchecked infrastructure spending.

Apple Slides After Supply Shortages Hurt Sales Forecast · Bloomberg Apple reported a 22% jump in iPhone sales for its third quarter, defying expectations as consumers actively upgraded their devices. However, the company’s stock slid in late trading after executives warned that severe component shortages, exacerbated by the global AI memory boom, will heavily constrain future hardware sales.

Amazon’s Zoox Wins First US Approval For Paid Robotaxis Without Human Controls · Slashdot Amazon-owned Zoox received a first-of-its-kind exemption from the National Highway Traffic Safety Administration to deploy and charge for rides in vehicles built entirely without steering wheels or pedals. The approval allows Zoox to commercially deploy up to 2,500 of its custom-built robotaxis annually, beginning operations in Las Vegas.

Google DeepMind’s new AI model can control a robot’s entire body · The Verge Google DeepMind introduced Gemini Robotics 2, an AI model that advances from upper-body control to managing “whole-body motions” for humanoid machines. By combining vision, language, and action models, the system enables robots to autonomously navigate, crouch, and manipulate physical objects in dynamic environments.

Who Wins and Who Loses After US Bans Foreign Robots? · Slashdot A sweeping new FCC ban on foreign-made mobile robots—covering everything from affordable Chinese research quadrupeds to household robot vacuums—has sparked intense industry debate. While proponents view it as a necessary defense of the US robotics ecosystem, critics warn it will severely handicap domestic startups and researchers by cutting off access to low-cost hardware platforms.

LinkedIn actually adds a ‘seems like AI slop’ button · The Verge In a blunt acknowledgment of platform degradation, LinkedIn is rolling out a reporting feature that allows users to flag posts as “Seems like AI slop”. The defensive move comes after third-party data showed that over 40% of the platform’s longform posts are now entirely generated by artificial intelligence.

Also Worth Knowing#


Categories: News, Tech