MOZOM vergelijkt
MOZOM compares: AI access, cybersecurity or political eligibility list?
- Source
- MOZOM vergelijkt
- MOZOM headline
- MOZOM compares: AI access, cybersecurity or political eligibility list?
- Original headline
- OpenAI and Anthropic restrict access to new powerful AI models to government-approved customers during a cybersecurity review
- Author
- MOZOM-redactie
- Date
- 27 juni 2026 om 13:32
- Subject
- New AI models only become available to approved customers during a cybersecurity review, causing security and political access to become intertwined.
Summary of the original report
According to AP, OpenAI and Anthropic are restricting access to new, powerful AI models during a US cybersecurity review. Early access would only be available to a limited group of government-approved customers. The official frame is security: first testing whether models can be misused for cyber attacks, vulnerabilities or national risks. The critical frame is access power: if the state determines who can use a model, a political admission list is created around a technology that is becoming increasingly important economically and militarily.
Striking in this message
It is striking that AI is no longer treated here as an ordinary product, but as strategic infrastructure. A model is then not only launched, but accepted.
Less visible context
What remains less visible is that access to top models gives power: better automation, better cyber defense, better analysis and sometimes better offensive capabilities. A temporary review can therefore be a temporary policy, but it can also create a permanent hierarchy between parties that were given access and parties that had to wait.
Possible message behind the news
A possible message is that security does not remain neutral when access itself becomes the scarce resource.
Neutral conclusion
The neutral conclusion: Pre-testing AI models can be defensible. But once only approved customers participate, the debate shifts from security to power, competition and political access.