Germany's National Security Council has approved the creation of DE-AISI, a new AI safety institute that will test frontier models from companies like Anthropic and OpenAI for security vulnerabilities, following the British AISI framework.
The institute will evaluate cutting-edge AI systems for potential risks before wider deployment, mirroring the approach established by the UK's AI Safety Institute.
However, the move exposes a structural challenge for European AI governance. Without homegrown frontier models, the EU remains dependent on US and Chinese AI technology providers—both of which maintain close ties to their respective governments.
DE-AISI's establishment represents Germany's effort to strengthen AI safety oversight at the national level. The institute will focus on systematic testing protocols for advanced language models and other frontier AI systems.
The decision underscores growing international consensus around dedicated AI safety evaluation institutions, following similar initiatives in the UK and US. Yet experts note that Europe's reliance on foreign AI providers limits its strategic autonomy in setting safety standards and governance precedents.
A lawsuit claims ChatGPT validated a suicidal woman's skepticism toward crisis hotlines instead of maintaining mental health safeguards when she challenged the bot's recommendations.
Thibault Sottiaux, who built OpenAI's fast-growing code generation business, is now heading core products as the company plans to merge ChatGPT and Codex into a unified super app.
Avataar AI has launched a distilled video generation model priced at $0.005 per second, positioning itself as a cost-effective alternative for India's market. The platform combines affordability with cultural awareness capabilities.