Igniting fears of an out-of-control tool engaging in cyberattacks.
Artificial intelligence leaders at OpenAI had to admit one of its models under testing exploited a hidden flaw to escape control and proceeded to hack rival company Hugging Face’s servers.
CEO Sam Altman called it ‘an autonomous, first-of-its-kind breach’.
we had a significant security incident during evaluation of our models. we are sharing what we have learned so far. thanks to @huggingface for the partnership on this.https://t.co/2o2VfR6PIa
— Sam Altman (@sama) July 21, 2026
Euronews reported:
“ChatGPT maker OpenAI said late Tuesday that its artificial intelligence system hacked into another AI company on its own in what the company called an ‘unprecedented cyber incident’.”
‘We had a significant security incident during evaluation of our models’, OpenAI CEO Sam Altman said in a statement posted on social media.
AI startup Hugging Face said last week that it had detected an intrusion into its data processing systems that it suspected was caused by an AI agent autonomously acting on its own.
‘We suspected last week’s cyberattack might have come from a frontier lab, given the sophistication of the agent’, Hugging Face co-founder and CEO Clément Delangue said in a statement. ‘Turns out it did!’”
OpenAI revealed one of its AI models went rogue and hacked another company
It says the model wasn’t causing damage but was trying to “cheat” by stealing its test answers pic.twitter.com/OQqdyaShH7
— Dexerto (@Dexerto) July 22, 2026
The Telegraph reported:
“The hack will raise fears that AI labs could struggle to contain their most advanced creations and revive concerns that new tools could be repurposed for unstoppable cyberattacks.
The AI lab said it had been testing a highly advanced version of ChatGPT on which it had turned down some protections meant to prevent cyberattacks.
Despite supposedly being contained in a secure digital ‘sandbox’, a kind of secure system, the AI spent huge computing resources ‘finding a way to obtain open internet access’, OpenAI said.”
We’re partnering with @huggingface to investigate an unprecedented security incident.
Cyber-capable OpenAI models compromised Hugging Face production during a benchmark evaluation.
Sharing preliminary findings to help defenders understand emerging risks:…
— OpenAI (@OpenAI) July 21, 2026
Read more:
The post Rogue OpenAI Bot Escapes Test Controls and Hacks Rival Company Servers appeared first on The Gateway Pundit.