TikTok parent company fires intern after an AI sabotage
AI may take our jobs, but first, AI could cost us our internship. That's what allegedly happened to an unnamed student who was part of a big AI project by ByteDance, TikTok's parent company.
ByteDance recently shed light on a surprising incident that sparked quite a buzz across Chinese social media platforms. South China Morning Post reports that ByteDance confirmed it terminated an intern back in August for what it described as "malicious interference" with an internal AI model training project. ByteDance clarified that the disruption was limited in scope and that reports online, which speculated about serious damage, were more fiction than fact.
According to ByteDance, the intern, a member of their commercialization tech team focused on ad technology development, had no direct connection to ByteDance's advanced AI projects. The incident, ByteDance stressed, had no bearing on the company’s main commercial ventures, online services, or its large language AI models, and certainly did not involve the intern in its renowned AI Lab. The company has also flagged the misconduct to industry associations and the intern’s university, with further steps potentially in the works.
This clarification comes as ByteDance has been investing heavily in generative AI, competing with OpenAI’s ChatGPT and leading China's market with its AI model, Doubao. Also, earlier this month, the company unveiled "Ola Friend": a new open-ear wearable that enables users to chat with ByteDance's AI without even picking up their phones.
Doubao has become a major hit, boasting 47 million monthly active users in September – far outpacing Baidu’s Ernie Bot, recently rebranded as Wenxiaoyan, and Moonshot AI’s Kimi, with 12 million and 7 million users respectively.
To add some fuel to the fire, an anonymous GitHub user under the handle “JusticeFighter110” recently uploaded an audio clip alleging the intern, reportedly an undergrad at Beihang University and a master’s student at Peking University, had confessed to planting malicious code". Another repository quickly emerged claiming the audio was a fake.
ByteDance responded, noting that while the intern had previous experience with top-tier organizations like SenseTime and the University of Oxford’s Torr Vision Group, some details shared on LinkedIn and GitHub about the intern’s experience at ByteDance were either misleading or inaccurate.
ByteDance recently shed light on a surprising incident that sparked quite a buzz across Chinese social media platforms. South China Morning Post reports that ByteDance confirmed it terminated an intern back in August for what it described as "malicious interference" with an internal AI model training project. ByteDance clarified that the disruption was limited in scope and that reports online, which speculated about serious damage, were more fiction than fact.
Speculation swirled on social media about the intern's motives, with some posts claiming the disruption was an act of protest over resource allocation, allegedly targeting a training session using a cluster of over eight thousand GPUs (video cards) – a hefty setup valued at millions. ByteDance was quick to call these claims overblown, asserting that the event was nowhere near as catastrophic as suggested.
This clarification comes as ByteDance has been investing heavily in generative AI, competing with OpenAI’s ChatGPT and leading China's market with its AI model, Doubao. Also, earlier this month, the company unveiled "Ola Friend": a new open-ear wearable that enables users to chat with ByteDance's AI without even picking up their phones.
To add some fuel to the fire, an anonymous GitHub user under the handle “JusticeFighter110” recently uploaded an audio clip alleging the intern, reportedly an undergrad at Beihang University and a master’s student at Peking University, had confessed to planting malicious code". Another repository quickly emerged claiming the audio was a fake.
Things that are NOT allowed: