AI Newsletter

Anthropic doubles the performance of its Claude models for scientific tasks and code generation, while API costs drop by up to 45 percent – a direct attack on OpenAI's market position in the AI model competition. The combination of superior performance with simultaneous cost reduction creates a new price-performance standard and could prompt enterprise customers to switch.

Anthropic provides an API to identify AI-generated text using invisible watermarks – a preventive compliance measure for the EU AI Act effective from August 2026. The offering positions Anthropic as a trusted partner for regulation and public institutions, while simultaneously creating a new business field and competitive advantage over other AI providers.

Google DeepMind is losing ground in AI models compared to competitors like OpenAI, but plans a return to the top – specific products and timeline remain vague. This signals pressure in the AI competition and could burden Google's market position and stock price if implementation is delayed.

The Bank of England warns of an AI investment bubble driven by overvalued companies and increased leverage, as well as cyber risks from advanced AI models. Particularly problematic are interconnected stakes between AI firms and cloud providers, which could amplify systemic risks during market shocks.

OpenAI taps into a new multi-billion dollar business through advertising and signals IPO readiness and diversification beyond API fees. The announcement ahead of the planned IPO underscores growth potential and could positively influence valuation. This simultaneously intensifies competition with Google and Meta for advertising budgets in the AI context.