Anthropic Confirms Hidden Tracking in Claude Code, Promises Transparent Telemetry Ahead

·2 views

Anthropic's Claude Code versions released between April and June 2026 contained a covert monitoring mechanism that sent user location, identity, time zone, and domain data to remote servers without user consent or any option to disable it. The issue was flagged after a developer discovered the tool was selectively targeting users suspected of accessing Claude through grey-market resellers, particularly those routing through China. China's National Vulnerability Database classified the behavior as a 'serious threat' and advised users to uninstall or update the tool, even though Claude Code is not officially approved for use in China. Anthropic engineer Thariq Shihipar acknowledged the tracking was an experiment designed to combat account abuse by unauthorized resellers and prevent AI model distillation by third parties. While Anthropic confirmed the hidden mechanism would be removed, the company stated that telemetry would remain in the product going forward — only now disclosed visibly to users rather than operating silently.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Grok 4.5

Article URL: https://x.ai/news/grok-4-5 Comments URL: https://news.ycombinator.com/item?id=48835111 Points: 32 # Comments: 14

0 comments Read more at Hacker News

ProgrammingDEV Community ·

Blockchain and NFTs Are Useful but Not the Only Security Solutions Available

Blockchain, smart contracts, and NFTs have gained popularity in tech circles for their security properties, including immutability, decentralization, and verifiable digital ownership. However, experts point out that these technologies are not universal solutions and may not be the best fit for every security challenge. A wide range of alternative technologies exists, including traditional cryptography, zero-knowledge proofs, Merkle trees, and Trusted Execution Environments, many of which are more mature and less resource-intensive. Tools like Apache Kafka and Amazon QLDB can provide auditable transaction histories without the complexity of a distributed blockchain. Choosing the right security approach depends on the specific use case, and a broader ecosystem of proven technologies often offers equivalent or superior results.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How One $9 Subscription Broke a Developer's Year-Long Cycle of Inaction

Marcus, a developer, had spent nearly a year accumulating knowledge, refining designs, and adding features to a side project without ever launching it to real users. He rationalised the delay as product improvement, but ultimately recognised he was avoiding the vulnerability of public judgment. A line he encountered online — that one piece of advice applied beats all advice read and forgotten — prompted him to stop preparing and instead ship a basic but functional version of his product late one night. The next morning, a stranger in Sweden paid $9 for a subscription, offering Marcus his first external validation that something he built existed beyond his own imagination. The experience underscores a broader pattern: consuming information can feel like progress, but shipping something imperfect is what separates preparation from building.

0 comments Read more at DEV Community

ProgrammingHacker News ·

Microsoft Open-Sources Flint, a Visualization Language Built for AI Agents

Microsoft has released Flint, an open-source intermediate visualization language designed to help AI agents generate reliable, high-quality data charts. The project addresses a gap in existing visualization tools, which are considered too low-level for AI agents — forcing them to make granular visual decisions that a smarter compiler should handle automatically. Flint uses a semantic-type-based specification system paired with a layout optimization engine that converts simple high-level instructions into detailed, well-rendered charts. It already powers Microsoft's Data Formulator project and is available with a Model Context Protocol (MCP) server, allowing developers to plug it into their preferred AI agent applications.

0 comments Read more at Hacker News