The Information Machine
ConcludedRan 10 days4 sourcesupdated 14 Nov 2026

DeepSeek's case for open-source AGI

The gist

Latest: DeepSeek Engineer Defends Open-Weight AGI Push Despite Safety Risks

A named engineer at a prominent Chinese AI lab has publicly stated the company will accept higher safety risks to speed AGI development. The remarks acknowledge the loss-of-human-control concern raised by AI researchers while explicitly prioritizing speed.

When stories like this move, I write about them in an email each morning.

One free email each morning, around 8 AM US Eastern.

The full picture

DeepSeek research engineer Gao Xi, in a rare public appearance at the 2026 World Internet Conference Wuzhen Summit, defended the company's policy of continuing to release model weights even while acknowledging higher safety risks. Gao said the goal is to accelerate AGI timelines, compressing a ten-year horizon to one year and then to a single day. Gao argued that whichever company achieves AGI first, the resulting benefits, such as healthcare breakthroughs, will accrue to all of humanity, and that open-source publication rather than closed development is the mechanism for getting there. Gao framed DeepSeek's competition as being against time itself, not against rival companies. South China Morning Post noted that some AI pioneers have warned AGI may escape human control once developed. The 2026 Wuzhen Summit included 'open source' in its theme for the first time, with figures including Joe Tsai and Liu Da Yiheng also endorsing open-source development.

How it developed
5 November 2026added 14 Nov

At the 2026 World Internet Conference Wuzhen Summit, DeepSeek engineer Gao Xi defended releasing model weights to hasten AGI despite higher safety risks, saying the company competes against "not other companies, but time".

Gao argued AGI benefits will accrue to all humanity regardless of which company produces it, with DeepSeek aiming to compress a ten-year horizon to a single day. South China Morning Post noted warnings from AI pioneers that AGI may escape human control once developed. The summit included 'open source' in its theme for the first time, with Joe Tsai and Liu Da Yiheng also endorsing open-source development.

Sources

Get the daily email

When stories like this move, I write about them in an email each morning.

One free email each morning, around 8 AM US Eastern.

See the latest edition · Privacy