BackChinese Think Tank Seeks Access to Anthropic's AI Model Mythos, Met with Refusal and US Alarm

En développement

Times of India13.05.2026TechIndia

Chinese Think Tank Seeks Access to Anthropic's AI Model Mythos, Met with Refusal and US Alarm

Mythos Anthropic AI Security

L'essentiel

A Chinese think tank's request for access to Anthropic's powerful AI model Mythos, capable of finding software vulnerabilities, was refused.
The US National Security Council reacted with alarm, highlighting concerns ahead of the Trump-Xi meeting in Beijing, where AI safety is now on the agenda.

Résumé généré par IA

Taille de police

A Chinese think tank's request for access to Anthropic's powerful AI model Mythos, capable of finding software vulnerabilities, was refused. The US National Security Council reacted with alarm, highlighting concerns ahead of the Trump-Xi meeting in Beijing, where AI safety is now on the agenda.

Read the full article on Times of India

Sujets liés

Mythos

SujetsAnthropic AI Security US-China Relations Trump-Xi Meeting

This article was originally published by Times of India.

Articles liés

Plus sur ce sujetMythos

En développement·7 sa önce

Chinese AI Models Gain Traction in US Amid Rising Costs

US companies are increasingly adopting Chinese AI models due to significant cost savings compared to domestic rivals. Models from DeepSeek and Z.ai are closing the performance gap, with Chinese AI usage via OpenRouter surging from 11% to 46% weekly.

Times of India

En développement·8 sa önce

First Diamond Artwork to Travel to Space on India's Vikram-1 Rocket

A specially engineered diamond artwork, Cosmic Bloom, will be the first of its kind to travel to space aboard India's Vikram-1 rocket by Skyroot Aerospace. The mission will test the diamond's performance in space, reflecting India's growing private space industry.

TOI World

En développement·9 sa önce

Orbital Seeks FCC Approval for 100,000 AI Data Satellites

Los Angeles startup Orbital applied to the FCC to launch up to 100,000 satellites, each an AI data center, into low Earth orbit. The plan aims to provide 10 gigawatts of computing power, with a demonstration mission next year and first dedicated satellite in 2028.

TOI World

En développement·9 sa önce

China Discusses Restricting Overseas Access to Advanced AI Models

Chinese authorities are reportedly in discussions with major tech firms to restrict overseas access to advanced AI models, citing national security concerns amid global competition. The proposed measures could impact the global AI market by limiting access to cost-effective models.

Times of India

En développement·13 sa önce

CISA Uses Anthropic's AI Mythos for Vulnerability Scanning Amidst US Policy Contradiction

The Cybersecurity and Infrastructure Security Agency (CISA) is reportedly using Anthropic's AI model Mythos to scan government software for vulnerabilities, despite ongoing disputes between the Pentagon and Anthropic over AI safeguards. The scans have already identified numerous bugs.

Times of India

En développement·16 sa önce

Andrew Ng Explains "Loop Engineering" in AI Software Development

Andrew Ng introduces "loop engineering," a new approach to AI software development. He details three core loops: the Agentic Coding Loop where AI writes and tests code autonomously, the Developer Feedback Loop where humans guide AI, and the External Feedback Loop involving real-world user input. Ng emphasizes the continued human advantage in providing context and vision.

Times of India

Plus sur ce sujetMythos