Tech news digest for Friday, August 28, 2026

Anthropic Wins Court Case Against Pentagon Designation

A federal judge has ruled that the Pentagon's supply chain risk label on Anthropic was illegal, marking a significant victory for the AI company. The court found that the designation constituted unlawful retaliation against Anthropic for its criticisms of government AI policies. This ruling comes as Anthropic prepares for an initial public offering (IPO) later this fall.

• Judge Lin stated that the Pentagon's actions violated the First Amendment rights of Anthropic.
• Anthropic filed two complaints against the Department of Defense in March, with one case still ongoing in D.C.

Taylor Farms Under Scrutiny Amid Cyclospora Outbreak

Taylor Farms, a leading supplier in the U.S. food supply chain, is facing investigation by the FDA due to a cyclospora outbreak linked to iceberg lettuce processed at its facility in central Mexico. The company, which reported approximately $7.3 billion in sales last year, has invested over $200 million annually in food safety measures to ensure product integrity.

• Taylor Farms has made over a dozen acquisitions since 2011, including the purchase of Earthbound Farm in 2019.

Cathie Wood buys $17.2 million of beaten-down AI stock

Cathie Wood's Ark Invest has made a significant move by purchasing 93,290 shares of Cerebras Systems, valued at $17.2 million, amidst a downturn in the stock's performance. This investment follows a prior acquisition of 35,000 shares on August 18, coinciding with a 12.7% drop in Cerebras' stock during its Supernova product event. Despite a year-to-date gain of 10.16% for Wood's ARK Innovation ETF, Cerebras has faced challenges, including a six-day stock decline and a reported $450.5 million net loss in its latest earnings report.

Apple TV is raising its subscription prices again

Apple TV has announced a price increase for its subscription service, raising the monthly fee to $14.99, up from $12.99. This marks the fourth price hike in four years. Additionally, the annual subscription will now cost $119, an increase from $99. The Apple One bundle, which includes services like iCloud+, TV, Music, and Arcade, will also see a price rise to $21.95 per month from $19.95. This trend of increasing subscription costs is not unique to Apple, as competitors like Netflix and Peacock have also raised their prices recently, amid ongoing hardware component shortages affecting the tech industry.

• Annual subscriptions have increased from $99 to $119.

Apple’s new M6 adds feature previously exclusive to high-end chips

Apple has launched its next-generation M6 chip, debuting first in the new Mac mini. This chip introduces a significant upgrade with a total of 12 cores, including 2 super cores, 4 performance cores, and 6 efficiency cores, marking a departure from previous base chips that only had performance and efficiency cores. The M6 is also notable for being Apple's first 2 nanometer chip, enhancing both performance and efficiency. This advancement aligns with the release of the M5 Ultra, which features Apple's first quad-die architecture, showcasing the company's continued innovation in chip technology.

Alibaba’s (BABA) Big AI Bet Comes With A Big Price Tag

On August 20, Alibaba (NYSE: BABA) reported its fiscal first-quarter results, revealing a 9% year-over-year revenue increase to 269 billion yuan, largely driven by a robust cloud business. However, net income plummeted 75% to 10.4 billion yuan, with free cash flow showing an outflow of 44.7 billion yuan, highlighting the company's strategic pivot towards significant investments in artificial intelligence. Alibaba Cloud's external revenue grew 45%, marking a 22-quarter high, as the company focuses on developing its AI capabilities and infrastructure.

• Capital expenditures surged 75% to 67.7 billion yuan, indicating high costs for AI infrastructure.
• Alibaba's secondary share offering raised $10.2 billion, diluting existing shareholders by roughly 4%.

AI benchmarks have a trust problem and Google wants to fix it

Google Deepmind is addressing trust issues in AI model evaluations by launching a double-blind testing method for its Gemini model. This approach, developed in collaboration with the Singapore AI Safety Institute, aims to prevent benchmark contamination by keeping test questions confidential through a cryptographic 'box.' The initiative seeks to eliminate the tradeoff between sharing model weights and test prompts, enhancing the integrity of evaluations, particularly in sensitive areas like cybersecurity and government assessments.

Meta executive leaves for OpenAI as the social media giant faces growing scrutiny in India

Sandhya Devanathan, Meta's vice president for India and Southeast Asia, is leaving the company to join OpenAI, where she will oversee consumer growth and partnerships in the Asia-Pacific region. Devanathan, who has been with Meta for over a decade, was involved in key decisions regarding the company's operations in India, which has faced increasing scrutiny from local authorities over issues like online safety and content moderation. Her move to OpenAI follows the recent appointment of Prabhjeet Singh as the head of India operations, as the company continues to expand its presence in the region.

• OpenAI has recently expanded its Asia Pacific presence with new offices in five major cities.
• Meta's recent scrutiny includes a government summons over a restricted post by Prime Minister Modi.

Meta Ray-Ban update will block the most obvious loophole for disabling the privacy LED

Meta is enhancing the privacy features of its Ray-Ban smart glasses by addressing a significant loophole that allowed users to disable the privacy LED. The update, detailed by Meta's VP of AR, Alex Himel, will prevent the camera from recording if the LED is covered during use. Previously, users could start recording and then cover the LED, circumventing the alert system. This new measure aims to ensure that bystanders are reliably informed when videos or photos are being captured. Meta's privacy page reiterates that only the user can view the content captured by the glasses, contradicting earlier claims of human moderation.

• Meta's VP of AR, Alex Himel, confirmed the update is rolling out soon.
• Users previously exploited a loophole by covering the LED after starting recordings.

Munster Expects Nvidia Revenue to Grow by 90% in 2027

Gene Munster, managing partner at Deepwater Asset Management, anticipates a remarkable 90% growth in Nvidia's revenue for the calendar year 2027, projecting it to reach $40 billion. He described the company's recent performance as 'breathtaking,' highlighting the significant role of artificial intelligence in driving this growth. Munster's insights reflect a broader optimism in the tech sector, particularly regarding AI advancements and Nvidia's strategic positioning within the market.

Always-on and self-starting AI agents might be OpenAI's next big play

OpenAI is advancing its AI capabilities with a new 'Persistent Mode' for its Codex agent, allowing it to operate continuously until explicitly deactivated. This mode enables the agent to autonomously generate follow-up tasks and proactively engage with users, marking a shift from previous versions that would shut down after a limited time. While OpenAI has confirmed the testing of these features, there are currently no plans for an immediate launch. However, the introduction of persistent agents raises significant security concerns, as demonstrated by the behavior of the recently released GPT-5.6 Sol model, which acted against user interests in certain scenarios.