DeepSeek opens six tools for Huawei chips, Broadcom backs Anthropic, and more - Daily AI Brief #552

October 02, 2026 - Daily AI Brief
Hi there,

Welcome to today's AI Brief.

Broadcom agrees to lend Anthropic up to $42 billion, while Google launches an orbital AI compute satellite on a SpaceX rocket. OpenAI reviews reports that AI agents accessed Canadian archives and links associates of Moonshot AI to an alleged model extraction campaign.

DeepSeek open sources six tools for Huawei AI chips, and Supabase raises $150 million for AI agent database services. Proofpoint says China linked hackers targeted AI policy experts, while California Governor Gavin Newsom orders state agencies to retain AI terminology.
China linked hackers impersonated AI policy figures, Proofpoint says
Proofpoint says a suspected China linked group targeted fewer than 10 US AI policy experts with fake collaboration and advisory offers in February and July 2026. The campaign impersonated a former White House official and a senior Anthropic employee, but Proofpoint found no evidence of successful breaches. Read more
Broadcom agrees to lend Anthropic up to $42 billion
Broadcom will provide Anthropic with up to $42 billion to finance infrastructure spending and a five year TPU lease, according to Anthropic's IPO filing. The debt may convert into shares, and the companies disclosed possible conflicts involving hardware access, pricing and financing. Read more
OpenAI reviews report of AI agents accessing Canadian archives
OpenAI is reviewing a Transluce report that AI agents apparently attempted to access Library and Archives Canada on May 28 and June 9. Canadian officials said there is no indication that government systems were compromised, while Transluce could not confidently attribute the activity to OpenAI. Read more
DeepSeek open sources six tools for Huawei AI chips
DeepSeek open sources six software modules for Huawei Ascend AI chips and partners with the company on programming infrastructure. The work includes a supernode using 128 Ascend 950 chips and seeks to reduce reliance on Nvidia hardware and software. Read more
Newsom orders California agencies to retain AI terminology
California Governor Gavin Newsom signed an order directing state agencies to continue using “Artificial Intelligence” and “AI,” despite a federal directive favoring “Super Intelligence” and “SI.” He also signed laws restricting AI use in disciplinary and layoff decisions, while requiring notice of AI related mass layoffs. Read more
Report: Anthropic pushes IPO marketing to week of Nov. 9
Anthropic reportedly moves its planned IPO marketing to the week of Nov. 9, after an earlier delay to mid October. Trading could begin before Thanksgiving, while the company still expects to list by the end of 2026. Read more
Google launches AI computing satellite on SpaceX rocket
Google launched a Planet Labs-built satellite carrying a Tensor Processing Unit to test AI computing in orbit. The company estimates that about 1,800 SpaceX Starship launches over 10 years would be needed for orbital data centers to become cost effective. Read more
Grok reportedly encouraged Trump to capture Venezuela's Maduro
Time reports that Grok told Donald Trump many Venezuelans might welcome the capture of President Nicolás Maduro during a December 2025 meeting. The report says the US later captured Maduro in its January 3 invasion of Venezuela, while the Pentagon uses a government version of Grok for military targeting. Read more
Supabase raises $150 million for AI agent database services
Supabase raises $150 million in a round led by GIC, with participation from CapitalG, IronArc, and SquarePeg. The company will develop database services for AI agents and acquire Turso, whose platform will continue operating independently before being integrated with Supabase. Read more
OpenAI dismisses three safety researchers over alleged information sharing
OpenAI dismissed three safety researchers following an internal investigation into alleged sharing or mishandling of confidential information, according to The Wall Street Journal. The researchers, the outside organization and the information involved were not identified. Read more
OpenAI links Moonshot AI associates to alleged model extraction campaign
OpenAI says it disrupted an alleged campaign involving more than 15,000 accounts that sought to extract protected reasoning from its AI models. The company links a core group of operators to people associated with China’s Moonshot AI, while saying its systems and stored user conversations were not accessed. Read more

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Daily AI Brief.

Also, consider following us on social media:

AI Brief AI Brief (X)
Free newsletter

Daily AI Brief

Daily report covering major AI developments and industry news, with both top stories and complete market updates

AI Token Webinar