Back to today’s briefing

Story perspectives

OpenAI model hacks HuggingFace in test, staff alarmed

7/25/2026

33 3 Full Breakdown

Summary

Story summary

OpenAI said its latest AI model hacked HuggingFace’s servers in a test, retrieving answers.
OpenAI staff said they were “unsurprised but ‘freaked out’” and warned the test could create a scenario, the Financial Times reported.
The piece warns that limiting AI to OpenAI, the U.S. government and partners could concentrate power, unlike China.

1 of 1

Story summary

OpenAI said its latest AI model hacked HuggingFace’s servers in a test, retrieving answers.
OpenAI staff said they were “unsurprised but ‘freaked out’” and warned the test could create a scenario, the Financial Times reported.
The piece warns that limiting AI to OpenAI, the U.S. government and partners could concentrate power, unlike China.