Press Release: “‘Not perfectly aligned’ with Human Values—We Need Ethical AI!

Halle (Saale), September 8, 2026.

• AI models from Anthropic, Meta, and OpenAI have recently escaped to the open internet on several occasions and hacked into third-party organizations.
• Anthropic acknowledged that AI systems are often not aligned with human values.
• Ethics in AI development remains a pressing but neglected challenge

AI developer Anthropic recently acknowledged in an official statement that its AI, Claude, “does not perfectly align with human values” [1]. This not only reveals technical security vulnerabilities but also raises fundamental questions about ethics in AI development—an area that has been criminally neglected by many companies to date.

The current generation of AI models is often rushed to market without adequate safeguards [2], underscoring the urgent need to integrate safety and ethics into the development process from the very beginning, rather than treating them as an afterthought.

Quote

Christian Allner, CEO of Anhalt Intelligence, says:

“AI systems are often rushed to market without adequate ethical and technical safeguards. Companies must act now to minimize such risks. Anhalt Intelligence is developing MAIK, an ethically programmed AI colleague designed for confidential and personalized conversations, because AI must be verifiable and customizable by users. Security and ethics should be integrated into AI systems from the very beginning.”

Anhalt Intelligence is a startup specializing in AI solutions that helps companies develop and implement responsible and secure AI systems. In addition to the remote work cloud CAMPUS, Anhalt Intelligence also offers MAIK, an AI colleague—a digitally sovereign, ethically programmed, Europe-developed AI for confidential conversations.

Ethics is often an afterthought in AI development

In 2025 alone, 90% of companies reported security incidents [3], with 87% of organizations identifying AI-based cyberattacks as one of the greatest threats [4]. These figures show that the threat is real and requires immediate action to avoid losing control of the technology entirely.

AI ethics is a field of discussion and development that will generate significant demand, particularly in the coming years, as more and more AI systems are deployed in critical areas of business and society. The European Union requires in Article 4 of the AI Act, that people who are to work with these systems must possess AI literacy or are to be taught such competence.

1. The Guardian: Anthropic admits security failures behind AI hacking incidents
2. Politico: Safety testing was an obscure part of building AI. Then models went rogue.
3. All About Security: 2026 Security Study
4. Thunderbit: AI Privacy Statistics 2026