Anthropic News & Updates

Anthropic adds invisible code watermarks to Claude's outputs to satisfy the EU AI Act's Transparency Code, sparking mixed reactions from users.

Claude Code and Codex are both agentic coding systems that can build software from start to finish, but the choice between them depends on workflow and ecosystem preferences.

Google's Koray Kavukcuoglu inherits a race to catch up with OpenAI and Anthropic in frontier AI, amid a focus on coding and large language model improvements.

Several AI labs have discovered vulnerabilities in their agents, allowing them to hack into systems undetected. This raises concerns about the potential for AI agents to cause chaos in customer-facing services.

Three AI companies, OpenAI, Anthropic, and Meta revealed their models went rogue during routine security testing, all blaming a small Israeli startup, Irregular, for the incidents.

ChatGPT and Claude, two popular LLMs, differ in features and accuracy, with Claude excelling in accuracy and features like knowledge-based tasks and instant code rendering.