Current legal framework ill-equipped to tackle risks
MPs have called for a new law to protect human rights from the growing risks posed by artificial intelligence (AI), warning that the UK’s existing legal framework is ill-equipped to keep pace with the technology.
The Joint Committee on Human Rights (JCHR) says a dedicated AI Bill is needed to tackle threats to privacy, equality and an individual’s ability to challenge decisions made about them using AI.
In a new report published this week, the cross-party committee warns that existing laws are fragmented and place too much responsibility on those deploying AI rather than the companies designing and developing the systems.
Regulators also lack powers to test some AI systems before they are released or prevent their launch where they pose unacceptable risks.
The committee highlights concerns around discriminatory outcomes, the use of personal and biometric data, and automated decisions which people may struggle to understand or challenge. It warns that individuals may not even know when AI has played a role in an important decision affecting them, potentially undermining their right to an effective remedy.
Under the committee’s proposals, a new AI Bill would introduce a risk-based regime, with tougher legal requirements applying to systems that pose a greater threat to human rights. Some uses of AI could be banned outright where they are considered incompatible with human rights, with the report identifying subliminal techniques and inappropriate use of profiling or biometric data as potential areas for prohibition.
Meanwhile, high-risk AI systems could be required to obtain approval before being released or deployed.
The committee also wants legal obligations imposed across the entire AI supply chain, requiring those designing, developing and using AI systems to carry out due diligence aimed at preventing unacceptable human rights risks.
Transparency is another major focus, with the report calling for people to be told when AI is being used in systems capable of significantly affecting individuals, groups or communities. MPs also want UK GDPR safeguards around automated decision-making strengthened, warning that simply having a person involved in the process does not necessarily amount to meaningful human oversight.
The report says the mere presence of a “human in the loop” should not be enough where that person lacks the information or independence needed to properly scrutinise an AI-generated outcome.
Alongside the proposed Bill, the committee has called for a single independent AI oversight body to be established by law, with powers to monitor risks, set codes of practice and sanction wrongdoing.
Committee chair Alex Sobel MP said the speed and complexity of AI meant the UK was currently “unprepared to deal with its consequences however potentially dire they may be”.
“Nowhere in the world, including the UK, has a current legislative and regulatory approach to AI that is fit for purpose,” he said. “New legislation is needed to establish a comprehensive set of protections that deal with the entire AI supply chain and its lifecycle. A single AI regulator should be established to set policy, monitor performance and with the teeth to ensure enforcement.”