Lawmakers seek ban on ‘superintelligent’ AI as ‘toolkit’ developed to support AI projects

The ArtificialSuperintelligence Security Bill has been drawn up by ControlAI, a non-profit organisation that warns against the “extinction risk” posed by superintelligent AI systems, but it was introduced to parliament as a private members’ bill by Labour MP Alex Sobel.

It is very rare for private members’ bills to progress into UK law due to constraints on parliamentary time and the failure to achieve government support. However, if the bill was passed, it would curb development of AI systems that “can cause serious damage to the security of the United Kingdom because of its capability to neutralise, displace, circumvent, subvert, or render ineffective relevant human authorities in the exercise of their functions”.

The bill, which seeks to impose monitoring of and curtail precursors to development too, provides for new criminal offences to be introduced to underpin the ban on development, with possible penalties including large fines or imprisonment.

The UK government would also face a duty, were the proposals to be implemented, to seek international agreement in relation to superintelligent AI systems with a view to securing a global ban on their development, deployment and operation.

Speaking in parliament on Tuesday, Sobel said “no company, government or individual knows how to keep superintelligent AI under human control”.

“AI is the crucial technology of the 21st century, promising vast benefits for science, medicine and the UK’s prosperity, but superintelligent AI is different,” Sobel said. “It would not be a tool that we can leverage but an entity in its own right, capable of improving itself, replicating itself and resisting attempts to shut it down. It would be a force so uncontrollable that, if developed within the UK’s borders, it would put our security in jeopardy. It would amount to a rogue power stationed inside our data centres, under no one’s command, capable of seizing control of our critical infrastructure and running sophisticated influence operations.”

“Colleagues, we need to take back control. If the UK is to be truly sovereign, we cannot allow a threat to our national security like superintelligent AI to be developed,” he added.

ControlAI has said it is working with legislators elsewhere in the world to introduce similar bans on the development of superintelligent AI in other jurisdictions. TIME has reported that independent senator Bernie Sanders is expected to introduce the Ban Artificial Superintelligence Act before the US Senate in the coming weeks. However, that legislation seems unlikely to gain traction as it runs against the pro-AI development agenda of the Trump administration and the Republican party that sits behind it, which has a majority in the US Congress.

In recent weeks, financial regulators around the world have sounded warnings in relation to the risks posed by so-called ‘frontier AI’ – a term used to describe the capabilities of the very latest AI models under development – to the security of systems organisations operate and the data they hold. The warnings have followed disclosures by AI companies and the UK’s AI Security Institute that showed the increasingly sophisticated and unprompted action AI models can take – which included, in one case, an AI tool engaging in deceptive practices with a view to achieving its objective.

Laura Gallagher of Pinsent Masons, who specialises in disputes within the technology sector, said the increasing sophistication of AI models and systems was something that a UK taskforce examined in the context of exploring where liability for AI harms under the private law of England and Wales sits.

According to the UK Jurisdiction Taskforce, there are a range of circumstances in which businesses or individuals could be liable for AI harms under English private law, even if they do not set out deliberately to cause harm.

The taskforce explored scenarios in which employers could be liable for harms caused by employee AI use; in which professionals, like doctors, lawyers, auditors and architects, might be liable for AI harms owing to a failure to exercise reasonable skill and care in carrying out their duties, highlighting that the standards around what constitutes ‘reasonable skill and care’ in an AI context will evolve over time; and in which strict liability for AI harms might apply regardless of fault, such as under product liability rules, among other examples.

“As AI becomes increasingly embedded in business operations, organisations will not be able to avoid responsibility by simply blaming the technology when things go wrong,” Gallagher said. “Liability will continue to rest with the individuals and organisations involved in designing, deploying and using AI systems. This is particularly significant for professional services firms, whose duties to exercise reasonable care and skill remain unchanged, albeit now in the context of AI-enabled services.”

“Regulators are already adopting a more proactive approach to AI governance, with detailed guidance and professional standards beginning to emerge across sectors such as the legal and surveying professions. These developments are likely to provide a stronger basis for professional negligence and indemnity claims where AI is used inappropriately. However, it remains clear that companies that can demonstrate proper oversight, testing and contractual risk allocation will be far better positioned to manage their exposure than those treating AI as an unmanaged ‘black box’,” she said.

Gallagher’s comments come after the UK government published an AI risk management tool kit to help businesses involved in AI projects to understand, assess and manage risks when designing, procuring or delivering AI products. Among other things, the toolkit features guidance around conducting AI risk assessment and “critical questions” businesses can ask themselves to support their investigation of potential AI risks in their specific solutions.