Concordia AI

What it does

NOT a government body. A social enterprise 'dedicated to advancing global AI safety and governance', with offices in Beijing and Singapore. It is the de facto bridge organisation between the Chinese and international AI safety communities: it publishes the 'State of AI Safety in China' report series and the AI Safety in China newsletter, co-authored the Frontier AI Risk Management Framework with Shanghai AI Lab, and runs the Chinese Perspectives on AI Safety translation project. Included here because for most outsiders it is the only functioning entry point to Chinese AI safety governance.

What they ask for — in their own terms

Build AI safety capacity and international coordination in and with China — better risk monitoring, standards input and cross-border dialogue.

What they do

Research, convenings, consulting to AI labs on risk management, contributions to Chinese national AI standards and policy consultations, and frontier risk monitoring across cyber, bio, chemical and manipulation domains. Runs the AI Risk Monitor and the State of AI Safety in China reports.

Its powers

research

Honest assessment

The single most reliable route for a non-Chinese-speaker trying to understand or engage with Chinese AI governance. English-language, responsive, explicitly international in mission. The caveat is the obvious one: it is a private organisation with no authority, cannot transmit complaints to Chinese regulators, and its access depends on remaining uncontroversial with Chinese authorities. Use it for understanding, not for redress.

What a newcomer can do

Subscribe to the AI Safety in China newsletter (aisafetychina.substack.com) and read the State of AI Safety reports; apply for a job if you have relevant expertise.

Worth knowing

It is a social enterprise that consults for AI companies, so it is not a disinterested campaigner and takes no donations or volunteers. There is no advocacy route for a member of the public — its value to you is information.

Concerns it covers

Frontier AI risks including loss of control, CBRN and cyber misuse, framed through international coordination — with the specific thesis that US-China mutual understanding is a precondition for any effective global regime.

Tips

no

Where they work

China (Beijing) and Singapore, with global engagement.

Size

Small; headcount not disclosed. Note it describes itself as a 'social enterprise', not a charity.

Spending

Commercial consulting revenue; no donation route.

Notes

Category is marked 'government' only because the output schema requires that field; this organisation is explicitly non-governmental.