The AI Decode

A 15year old in Japan was arrested after allegedly using ChatGPT to automate an attack that unsubscribed 46,812 Bandai anime streaming users. The hack forced the service offline for six weeks and may have exposed data from up to 1.36 million accounts.

Alibaba bans Claude Code after Anthropic confirmed it included a hidden feature to detect unauthorized China linked access and prevent model theft. Employees have been told to switch to Alibaba's Qoder AI coding assistant.

Anthropic faces a $75 million lawsuit from over 100 authors who allege the company used their books without permission to train its Claude AI models.

ANTHROPIC CRACKS DOWN ON CHINESE ACCESS Anthropic is tightening restrictions to block users in China from accessing Claude AI, targeting VPNs, proxies, overseas accounts, and API resellers to enforce U.S. export rules.

JUST IN: SpaceX reportedly showed investors an early AI device prototype. It's said to be slimmer than an iPhone, powered by xAI, run a proprietary OS, and use a Qualcomm Snapdragon chip. The project is still in early development, with no guarantee it will launch.

JUST IN: AI chip startup Etched raises $800M and signs $1B in sales contracts, positioning itself as an NVIDIA challenger.

JUST IN: Meta bans employees from using competing AI models.

Both the U.S. and China have integrated AI into their nuclear early warning systems meaning an algorithm could help determine whether a nuclear war begins

JUST IN: Investor Marc Andreessen says "Doctor ChatGPT" is a better doctor than 99% of doctors.

JUST IN: Claude for Social Media Accounts is here. Simply enter your website, and Fastlane creates hundreds of TikTok and Instagram accounts, then warms them up in your niche using real human operators. This is seriously insane.

In a demo, Anthropic's Mythos model drained bank accounts and generated a detailed plan to kidnap a lawmaker in just 30 seconds.

Google has reportedly restricted Meta's access to Gemini amid capacity shortages, according to the Financial Times, delaying Meta's AI projects for customer support and content moderation.

IBM has unveiled the world's first sub 1nm chip technology, packing nearly 100 billion transistors into a fingernail sized chip. IBM says the new architecture could deliver 7,000 TOPS up to 7× faster than today's AI accelerators and reduce frontier AI model training from months.

TODAY: OpenAI may postpone its IPO until 2027 as it targets a $1 trillion valuation, with CEO Sam Altman reportedly rejecting any lower figure, according to NYT.

Apple has significantly raised prices due to increasing costs, with the fully loaded 16-inch MacBook Pro now priced at $9,999.

🚨 OpenAI has unveiled Jalapeño, its first custom AI chip, built with Broadcom and Celestica. Designed to power ChatGPT and future AI agents more efficiently, the chip is expected to begin deployment by the end of 2026. Read more for full details👇🏻 theaidecode.com/artificial-int…

JUST IN: The Pentagon confirms Grok AI helped launch more than 2,000 missiles at 2,000 targets across Iran in just 96 hours

Tesla owners will be able to join the Robotaxi network and earn up to 75% of the revenue, according to Elon Musk.

While everyone was watching Elon Musk’s AI1 satellite, China quietly made its own move. Beijing launched a Space Computing Innovation Center focused on AI chips, compute satellites, and space-based AI infrastructure—one week before AI1 grabbed headlines. 🚀🤖

Forget ChatGPT generate AI answers after the apocalypse. SqueezLabs built "CrankGPT," a hand-cranked AI box that runs completely off grid no batteries, no solar, just human power. Turn the crank, power the model, get answers.