OpenAI says it detected suspicious and potentially malicious activity months before a reported cyberattack targeting Hugging Face, an AI development platform.
The company said its security systems identified activity that raised concerns and that it had been monitoring the situation before the attack became public. The disclosure highlights the growing difficulty of protecting AI platforms and their users from sophisticated cyber threats.
The incident has also raised questions about how quickly organisations can detect and respond to malicious activity targeting AI infrastructure. Hugging Face hosts models, datasets and tools used by researchers and developers around the world, making security a major concern.
OpenAI's disclosure comes amid increasing warnings that attackers are using increasingly advanced techniques to target AI systems, developer platforms and the data associated with them.