Superintelligence is now a governance question for leaders

Superintelligence usually means AI that outperforms the best humans across most intellectual work. No lab has shown such a system. Lab leaders have claimed early versions could arrive within a few years, while researchers such as Yann LeCun and expert forecasting panels are more skeptical. ETA does not predict dates. We track measured capabilities, published evaluations and lab disclosures so leaders can judge progress on evidence rather than headlines.

It is a voluntary pledge signed on September 29, 2026 by leaders of Anthropic, Google, Meta, Nvidia, OpenAI and xAI. Signers commit to internal monitoring of model capabilities and alignment, a team that owns those controls, independent external auditors, and an independent committee reviewing the results. A companion executive order directs agencies to use the term super intelligence and asks for proposed legislative language defining it. The accord is not a law and carries no new penalties.

A June 2, 2026 executive order invites developers of covered frontier models to give government up to 30 days of access before broad release, without licensing. In October the president named DNI Jay Clayton to lead a Super Intelligence Force with 120 days to report on risks and opportunities. In November the Justice Department launched its own Super Intelligence Task Force to coordinate enforcement against AI misuse. Most measures so far are voluntary or security focused.

Focus on what you control. Ask AI vendors which safety commitments they have signed and how incidents are reported to customers. Scope AI agent permissions like privileged accounts and confirm monitoring cannot be quietly disabled. Shorten patch cycles, since AI is speeding up vulnerability discovery. Base workforce plans on measured capability gains and revisit them quarterly. Watch the Super Intelligence Force report, which could shape procurement and compliance expectations in 2027.