Anthropic

2 articles · 1–2

Anthropic12 Sept 2026

Anthropic Blocked the Prompt. But Did Its AI Safety System Prevent the Harm?

Anthropic says it blocked AI misuse, yet some damage had already happened. The real AI safety question is whether “blocked” means detection, disruption or prevention.

Read analysis

Anthropic30 Aug 2026

Claude Knows Your Favourite Lyrics. But Did Anthropic Steal Them?

Sony and Warner say Claude was trained on pirated lyrics. With damages of up to $150,000 per work, the case could reshape how AI companies document their training data.

Read analysis