The TikTok parent is expanding its focus on AI safety as it develops artificial intelligence models and applications for global markets.
ByteDance, the Chinese technology company behind TikTok, has established a dedicated artificial intelligence safety team within its research division, Seed, and launched an international recruitment drive for 10 researchers as it strengthens safeguards for increasingly capable AI models.
The newly formed Alignment & Preparedness team will focus on identifying potential risks associated with advanced AI systems, improving model reliability and developing safety measures for autonomous AI agents.
The recruitment drive covers positions in Beijing, Shanghai and San Jose, California, reflecting the company's plans to expand its AI safety research across China and the United States.
The development comes amid growing scrutiny of AI agents capable of independently accessing the internet, interacting with digital services and completing multi-step tasks with limited human intervention.
AI alignment refers to the development of systems that operate consistently with human objectives, instructions and safety requirements. The field has gained attention as AI models become capable of performing increasingly complex activities.
According to a report by Reuters, ByteDance's announcement represents an uncommon public commitment by a major Chinese AI developer to establishing a dedicated research function for frontier-model safety and preparedness.
Chinese state-backed newspaper The Paper reported that Seed was the first Chinese frontier AI laboratory to establish a standalone department focused on safety and alignment.
In a recruitment announcement published on its official WeChat account, Seed stated that AI safety and alignment were becoming essential capabilities for deploying advanced models at scale, rather than functioning solely as measures to mitigate risks.
The company said the new team would receive computing capacity and research resources comparable to those available to its frontier-model development teams.
Its research priorities will include integrating safety training throughout model development, conducting automated vulnerability assessments and developing tools to monitor and interpret AI model behaviour.
The team will also work on testing environments for autonomous AI agents and preparedness measures designed to address emerging risks associated with more advanced AI capabilities.
ByteDance plans to extend its safety research across model training, evaluation and deployment, rather than limiting assessments to the final stages of product development.
The company also intends to collaborate with universities and research institutions and seek independent evaluations of its AI systems.
The announcement comes as the United States and China continue discussions concerning AI safety safeguards, risk management and communication mechanisms for significant AI-related incidents, alongside competition in advanced AI development.
For ByteDance, the establishment of a dedicated safety team introduces a specialised research function within Seed as the company continues developing increasingly capable AI models.
The recruitment initiative also places greater emphasis on technical testing, external evaluation and risk preparedness as part of the company's broader AI development activities.
Disclaimer: This article may include information derived from interviews, press releases, public statements, research, company communications and other publicly available or third-party sources. Such material may be summarised, paraphrased or contextualised for journalistic and editorial purposes. All rights in third-party content remain with their respective owners.