ByteDance, the parent company of TikTok, has addressed reports that surfaced online in China last week, stating that interns at the company planted malicious code that impacted the training of one of the company’s AI models, resulting in significant damage.
According to ByteDance, this incident was caused by interns in the technology advertising development department, not the AI oversight department, inserting attack code used for internal research and development purposes. This issue did not affect the AI models deployed and utilized by current clients.
ByteDance also refutes claims that this disruption affected up to 8,000 GPUs, dismissing it as an exaggeration.
Ultimately, ByteDance revealed that this event occurred in August, and the interns involved have been dismissed, with warnings issued to universities and relevant authorities.
ByteDance boasts an AI chatbot for Chinese users called Doubao and has developed several other applications like Jimeng for creating videos from text inputs.
Source: BBC
TLDR: ByteDance clarifies intern involvement in malicious code incident, denying impact on AI models and refuting inflated GPU disruption claims. Interns have been dismissed, and notifications sent to universities and authorities. ByteDance continues to innovate with AI chatbots and video creation applications.
Leave a Comment