AI News Briefing — October 3: Apple moves to tighten Full Disk Access on the Mac, U.S. charges a $300M AI server smuggling scheme

Today's Headlines

Apple will tighten Full Disk Access on the Mac, saying AI agents substantially raise the risk
The Justice Department arrests a company owner charged with smuggling more than $300 million in AI servers to China
Anthropic commits $100 million to train 10,000 engineers who deploy AI inside companies by the end of 2027
Nvidia adds a 64GB DGX Spark starting at $4,999

Today's three stories are about who holds the keys to AI. Apple plans to narrow a path that lets apps, including AI agents, reach everything on a user's Mac, and the Justice Department charged a man with sending AI servers to China without the licenses they required.

Anthropic, meanwhile, launched a program to train the engineers who put AI into production inside companies, held to the same standard as its own staff.

Today's Top Three

Apple moves to tighten Full Disk Access on the Mac as AI agents grow more capable

Apple told developers on October 2 that it will add new controls to Full Disk Access, the macOS permission that gives an app reach across a user's entire system.

Apple said the permission exists so backup apps can work properly, and that it largely sidesteps the controls designed to protect users' private data. Some developers, the company said, are using it in ways that expose files, mail, messages and even browsing history without users' full knowledge and understanding. For communication apps, that can also compromise the privacy of the people users talk to.

Going forward, users who genuinely want to grant an app this level of access will be able to do so only through "very explicit user action." "As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially," Apple wrote. The company did not give a timeline.

According to TechCrunch and The Verge, the announcement followed a report by Inc. columnist Jason Aten that Meta's Muse AI app knew the contents of his private messages. Meta's spokesperson said reading Messages requires users to enable both Full Disk Access and the Messages connector.

The Justice Department charges a $300 million AI server smuggling scheme routed through Malaysia and Singapore

The Justice Department announced on October 1 the arrest of Greg Lui, 38, owner of Earthmade Computer, a technology company in City of Industry, California, on charges of smuggling more than $300 million in export-controlled servers to China.

The three-count indictment, returned on September 29, charges conspiracy to violate the Export Control Reform Act and the Export Administration Regulations, outbound smuggling, and conspiracy to commit money laundering. Prosecutors say Lui gave U.S. manufacturers false paperwork naming destinations that needed no export license, shipped the servers to Malaysia and Singapore, and then re-exported them to China.

In one example from the indictment, Lui ordered 27 servers from a U.S. manufacturer in January 2024 for about $7,614,000 and had them shipped from Los Angeles to Kuala Lumpur. In March 2024, a co-conspirator told a Malaysian government official by email that all 27 had been transshipped to a buyer in China. The conspiracy and money laundering counts each carry a maximum of 20 years in prison, and the smuggling count up to 10 years.

Ars Technica reports that the FBI identified Nvidia A100 and H100 GPUs among the hardware. The Justice Department's release refers to AI as "Super Intelligence (SI)," the term the administration has adopted.

Anthropic launches Claude Frontier Academy with $100 million to train 10,000 deployment engineers

Anthropic announced Claude Frontier Academy on October 2, a program to train the engineers who build AI into production systems inside companies.

Backed by a $100 million commitment, Anthropic aims to train 10,000 Frontier Deployed Engineers (FDEs) by the end of 2027. The first cohorts include engineers from Accenture, Bain, Capgemini, Commonwealth Bank of Australia, Deloitte, McKinsey, Morgan Stanley and Novo Nordisk.

The first program, the FDE Residency, follows the way doctors are trained. Engineers start with a multi-day in-person course built around a simulated enterprise deployment, from choosing the use case through security review to handover, and finish with a graded practical. Those who pass move into a 12-week residency leading a real Claude project at their own organization, then face a final assessment for the Claude Frontier Deployed Engineer credential. The first graduates are expected in early 2027.

Cohorts run in San Francisco, New York and London, and participation is by company nomination. Anthropic says professionals across 46,000 firms have already earned more than 175,000 Claude certifications through its Claude Partner Network.

Claude Frontier Academy: $100M to train 10,000 engineers (Anthropic)

Other Developments

Products

Nvidia is adding a 64GB configuration of DGX Spark, its compact AI development computer. Acer, ASUS, Dell, Gigabyte, HP and MSI will sell it from October 23, starting at $4,999. A single unit runs models of up to 100 billion parameters, and two linked units delivered up to 1.7 times the performance of one in Nvidia's Qwen 3.8 27B test. PC Watch reports that Nvidia raised the price of the 128GB Founders Edition to $6,950, citing memory supply constraints and rising costs. NVIDIA DGX Spark 64GB Gives Developers More Ways to Build and Scale Local AI (NVIDIA) / メモリ64GB版「DGX Spark」登場。4,999ドルで、128GB版は値上げ (PC Watch)
Nvidia raised the price of the Shield TV Pro, a streaming box whose current version debuted in 2019, to $299 starting October 2. A spokesperson told Ars Technica that the cost of components, including memory, has risen substantially across the industry. The 7-year-old Nvidia Shield TV is now $100 more expensive due to AI (Ars Technica)

Other

Pope Leo XIV wrote on X that "in this era of artificial intelligence, it is becoming urgent to distinguish human art from what machines produce," adding that there is an ontological difference between the two even before an aesthetic one, TechCrunch reports. Pope Leo XIV is not a fan of AI-generated art (TechCrunch)

Source: Selected by the editors from the AI news inbox (collected October 3, 2026; 30 items, 11 primary and 19 secondary).