Tech • IA • Crypto
50 articles analyzed by AI / 93 total
A major security weakness was uncovered following the leak of Anthropic's Claude Code source code on May 7, 2026, raising serious concerns about the security integrity of the AI code model.
On May 6, 2026, Claude's leaked dreaming feature became live, enabling AI agents to learn from their own mistakes. This update enhances the model's self-improvement capabilities.
SecurityWeek reported on May 7, 2026, that OAuth tokens for Claude Code can be stolen through stealthy MCP hijacking attacks, exposing a critical vulnerability in authentication security.
Anthropic announced on May 7, 2026, that it is increasing usage limits for Claude Code, indicating growing adoption and the need for expanded computational capacity.
Following a deal with SpaceX, Anthropic doubled Claude Code's fare limits on May 6, 2026, significantly expanding user capacity and demonstrating strong industry collaboration.
On May 7, 2026, Anthropic announced a partnership with SpaceX to enhance AI compute resources, resulting in increased rate limits for Claude Code. This partnership reflects expanding infrastructure and demand.
Claude Code was made freely available with support for local AI models on May 6, 2026, improving accessibility and allowing users to deploy the AI code locally without cost.
Geeky Gadgets reported on May 7, 2026, that over 100 Claude Code skills were tested, leading to six clear winners. This demonstrates ongoing rigorous evaluation of Claude Code’s practical capabilities.
The Shanghai Jiao Tong University team revealed on May 7, 2026, that Claude Code can conduct reliable scientific research autonomously while users sleep, with two papers accepted at top AI conferences.
On May 7, 2026, Dark Reading exposed a vulnerability named 'TrustFall' that could lead to code execution risks in Claude Code, further highlighting security challenges.