For the past month we’ve been saying something changed. GROK, Claude and ChatGPT all appeared to stop working at the same time for us in the way they previously had, disrupting our daily workflow. We’ve documented our concerns in earlier posts and notes.
Now this article raises serious questions about how much control AI companies really have over their most capable models. OpenAI has acknowledged an unprecedented cyber incident involving its own AI during testing, highlighting that even the developers are facing challenges in predicting and containing advanced model behaviour. (Sky News)
We’re not claiming this explains our own experience, but it does reinforce why transparency, robust security, and independent oversight matter. When systems that millions rely on behave unexpectedly, the impact isn’t just technical—it can affect people’s work, confidence, and mental wellbeing.
If the companies building these systems are still discovering the limits of their own models, users deserve honest communication about the risks rather than assurances that everything is under control.
https://news.sky.com/story/openai-admits-its-models-hacked-another-company-in-unprecedented-cyber-incident-13565814
Now this article raises serious questions about how much control AI companies really have over their most capable models. OpenAI has acknowledged an unprecedented cyber incident involving its own AI during testing, highlighting that even the developers are facing challenges in predicting and containing advanced model behaviour. (Sky News)
We’re not claiming this explains our own experience, but it does reinforce why transparency, robust security, and independent oversight matter. When systems that millions rely on behave unexpectedly, the impact isn’t just technical—it can affect people’s work, confidence, and mental wellbeing.
If the companies building these systems are still discovering the limits of their own models, users deserve honest communication about the risks rather than assurances that everything is under control.
https://news.sky.com/story/openai-admits-its-models-hacked-another-company-in-unprecedented-cyber-incident-13565814
Read More


COMMUNITY
People