WorkSeptember 16, 2026via Financial Times Technology
AI must not outrun safety controls, DeepMind co-founder warns
Why it matters
As frontier labs race toward AGI-scale models, a founding voice at DeepMind is calling for governance and safety frameworks to catch up. The launch of a new DeepMind Institute signals serious effort to embed safety thinking into capability development—but also reflects real anxiety about the pace.
Key signals
- Shane Legg (DeepMind co-founder) warning: AI capability development is outpacing safety controls
- New DeepMind Institute launching to explore AGI implications
- Growing industry calls for slowdown in AI development
- Safety-capability gap framed as policy/governance issue, not pure technical risk
- Shane Legg (DeepMind co-founder) warns AI must not outrun safety controls
- New DeepMind Institute launched to explore AGI implications
- Growing industry calls for AI slowdown
- Published September 16, 2026
- Source: Financial Times (opinion/commentary framing)
The hook
DeepMind co-founder warns the industry is moving faster than safety can keep up—and a new institute is being built to prove it.
Shane Legg speaks to FT as new DeepMind Institute launches to explore implications of artificial general intelligence, amid growing calls for industry slowdown