70+ UK Lawmakers Urge a Ban on Superintelligence

We have another very exciting update this week! Over 70 cross-party MPs and peers are calling on the UK Prime Minister to lead an international agreement prohibiting the development of superintelligent AI, while preserving strategic AI ambitions.

In a letter published by Alex Sobel MP yesterday, this coalition is urging the Government to adopt the Artificial Superintelligence Security Bill, which was introduced by Sobel in Parliament on Tuesday in collaboration with ControlAI.

The letter has been covered in The Guardian, and ControlAI’s US Executive Director Connor Leahy spoke about it on BBC News yesterday evening!

AI promises prosperity and vast benefits in science and defense. Superintelligent AI, however, would be so autonomous and capable that it would replace and outcompete humans across the board. We would not control it, and experts warn it poses an extinction threat to humanity.

Just this week, AI researcher Jacob Coxon quit Anthropic, warning that AI companies are “gambling with our lives” by racing to build self-improving superintelligent AI. Evan Hubinger, a top AI safety researcher at the company, commented on this and said that Coxon is correct, and he personally thinks there is a greater than 10% chance that AI will “kill all humans” within the next decade. Many researchers think the risk is even larger. Geoffrey Irving, who until recently was the Chief Scientist of the UK’s AI Security Institute, and has led safety teams at OpenAI and Google DeepMind, says he thinks there’s a 50% chance everyone dies as a result of superintelligence.

The Artificial Superintelligence Security Bill prohibits superintelligent AI in the UK. It strikes a balance to address the threat, while ensuring the UK can pursue its AI ambitions.

The way it achieves this is by empowering the Government to monitor and restrict precursors to superintelligence, such as advanced AI chips and AI capabilities like hacking.

Thus, the UK could detect and intervene on superintelligence development before any project gets dangerously close, while continuing to make use of precursors for legitimate purposes.

If developed within the UK’s borders, superintelligence would amount to a rogue army inside our data centers, under nobody’s command. A domestic ban prevents this.

To tackle the threat abroad, we need an international coalition to prevent it from being developed anywhere. The bill sets the policy for the UK to pursue an international agreement to prohibit superintelligent AI worldwide.

With its upcoming G20 presidency, Britain is well positioned to lead on this, as it did in hosting the first global AI Safety Summit at Bletchley Park.

The bill has been endorsed by renowned AI scientist Professor Stuart Russell, OpenAI whistleblower Daniel Kokotajlo, Nobel Peace Prize laureate Beatrice Fihn, and Sir Stephen Fry.

The new momentum behind the bill builds on our UK campaign, backed by 130+ lawmakers who recognize the threat. It’s great to see the coalition to prevent the threat posed by superintelligent AI grow.

The warnings have been heard. It is time to act.

We’ll be updating the full list on our website as more signatures come in here: https://controlai.org/uk-pm-letter