Anthropic filed an S-1 registration statement with the SEC for a proposed initial public offering of Class A common stock, planning to list on the Nasdaq Global Select Market under the symbol ANT.
The filing reveals H1 2026 revenue of $16.3 billion alongside a net loss of $80.7 billion. Anthropic does not expect to pay a dividend in the foreseeable future and warns that current revenue growth rates may not be sustainable.
Capital expenditure plans include up to $120.5 billion allocated for SpaceXAI and Nvidia compute, plus up to $114.5 billion committed to Google through 2033. The company notes it may need to seek additional financing sooner than expected.
An allocation of 5 percent of shares is designated for the Anthropic Public Benefit Fund. The registration statement also includes formal warnings regarding existential risks associated with advanced artificial intelligence.
Market expectations point to a fundraising target of up to $100 billion at a valuation near $2 trillion, with an IPO anticipated during the week of November 16th.
The DarkSword iOS exploit chain has moved from theoretical vulnerability research to a fully deployed operational toolkit targeting cryptocurrency wallets and keychain data.
The chain targets iOS versions 18.4 through 18.6.2, executing through Safari visits without requiring user interaction beyond opening a specific page. Following public disclosures of the underlying CVEs and source code reconstruction on GitHub earlier in the year, a complete deployment package appeared online. This package includes the exploit components, a delivery server, an administrative panel, a database, and operational documentation.
Public tracking indicates broader distribution vectors beyond standard watering hole attacks. App Store poisoning incidents identified embedded kernel exploit modules in applications such as FomoPeek and a food delivery application named ComeCome. These embedded components bypass device sandboxes to access local data, notes, and messaging applications.
Deployments associated with the administrative panels accept devices running iOS 13.0 and newer, leveraging extensive payload libraries. Security monitoring systems continue to track indicators of compromise across mobile application stores and malicious web payloads.
Dessa is a full duplex speech model trained from scratch exclusively on human conversation without text data. The system operates locally on Apple Silicon hardware and runs inference faster than real time.
block.xyz/dessa




















