STORY · FORSKNING_
Security expert: Open source models from 2025 can already hack networks
Thomas Ptacek, a recognized security researcher, believes that publicly available AI models from 2025 are capable of performing sandbox escapes and network hacking. He argues that this is unsurprising because OpenAI likely has better security measures than others.
WHY IT MATTERS
This underscores that AI security threats are no longer limited to frontier models from major labs – open models represent a genuine operational cybersecurity risk that businesses must take seriously.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.