Commons Sense

House of Commons · Ministerial Statement

AI Seoul Summit

23 May 2024 · 14 other contributors

Opened by Saqib Bhatti Con Meriden

Science & Technology

Summarised by AI from the official record, so it can contain mistakes.

Read the full transcript on Parliament.uk

At a glance

Saqib Bhatti raised concerns about ai seoul summit in the House of Commons. A government minister responded. Other MPs also contributed.

Key points

  • The Minister for Science, Research and Innovation announced the outcomes of the AI Seoul summit co-hosted by the UK and South Korea.
  • The event follows the first AI safety summit held at Bletchley Park in November 2023, where 28 countries signed a declaration to develop trustworthy AI.
  • Since the Bletchley Park summit, the UK has launched its AI Safety Institute and established an office for pre-deployment testing of advanced AI systems.

How the debate unfolded

MPs spoke in turn to share their views and ask questions. Here's what each person said.

Government Statement

Saqib Bhatti

The Minister for Science, Research and Innovation announced the outcomes of the AI Seoul summit co-hosted by the UK and South Korea. The event followed the first AI safety summit held at Bletchley Park in November 2023, where 28 countries signed a declaration agreeing to develop trustworthy artificial intelligence. Since then, the UK has launched its AI Safety Institute, which has conducted pre-deployment testing for potential harmful capabilities on advanced AI systems and established an office in San Francisco to broaden technical expertise. The Government also announced high-level partnerships with France, Singapore, and Canada. At Seoul, world leaders agreed to establish a global network of state-backed organisations dedicated to frontier AI safety, sharing information about models' limitations and risks. Sixteen leading companies signed commitments to enhance AI safety and not release new models if risks are too high. Countries also pledged to develop shared risk thresholds for model deployment, addressing severe risks like aiding malicious actors in acquiring weapons or evading human oversight. The statement highlighted the importance of supporting international co-operation on AI science and fostering an environment that nurtures easy access to AI-related resources for SMEs, start-ups, and academia.

Other contributors (14)
  • Chris Bryant Lab Rhondda

    Question

    How will shared risk thresholds be drawn up and what legal force will they have in the UK without legislation?

    Minister reply

    The minister did not provide a direct answer to this question, but generally emphasised the importance of international co-operation and the voluntary nature of commitments made at Seoul.

  • Chris Bryant Lab Rhondda

    Question

    Does the voluntary agreement from summit mean companies developing the most advanced AI are still marking their own homework despite potential risks?

    Minister reply

    The minister highlighted the importance of voluntary agreements and industry leadership, but did not directly address concerns about self-regulation.

  • Chris Bryant Lab Rhondda

    Question

    Which 'malicious actors' are referred to in the statement? Is this integrated with the cyber-security strategy?

    Minister reply

    The minister did not provide a specific answer but stressed the need for international collaboration on AI safety, which may be related to broader security strategies.

  • Chris Bryant Lab Rhondda

    Question

    Why have the Government not taken steps like President Biden's Executive order?

    Minister reply

    The minister highlighted the UK’s leadership in international AI safety efforts and did not directly address why specific measures like an executive order were not adopted.

  • Chris Bryant Lab Rhondda

    Question

    Have commitments made six months ago at the UK safety summit been kept?

    Minister reply

    The minister noted progress in AI safety since Bletchley, including the establishment of the AI Safety Institute and international partnerships, but did not directly address whether all previous commitments have been met.

  • Chris Bryant Lab Rhondda

    Question

    Why were no creative industries representatives present at the summit despite its importance?

    Minister reply

    The minister did not provide a direct response to this question but generally highlighted international and industry collaborations on AI safety.

  • Chris Bryant Lab Rhondda and Ogmore

    Question

    Expressed gratitude for advance sight of statement; paid tribute to Minister for Employment, Jo Churchill. Questioned how 'shared risk thresholds' will be drawn up, their legal force in the UK, whether companies developing advanced AI are marking their own homework, which malicious actors the Government refers to and how this work is integrated with cyber-security strategy. Professor Yoshua Bengio has said that more regulatory measures are needed but why have the Government not taken steps like President Biden’s Executive order? Have commitments made six months ago been kept or are these voluntary agreements just empty words? The code of practice working group has collapsed, what is the plan now and will legislation be considered?

    Minister reply

    Thanked Chris Bryant for his kind remarks about Jo Churchill. Emphasised AI safety as a priority; mentioned NHS benefits from AI technologies such as Brainomix. Stated that independent verification of AI models in healthcare is important. Addressed malicious actors through cross-government approach led by the Security Minister. Reiterated commitment to regulator-led approach and readiness to legislate when necessary.

  • John Hayes Con South Holland and The Deepings

    Question

    Complimented Government's lead on AI safety and establishment of the world’s first organisation dedicated to AI safety. Highlighted potential risks from underestimating AI effects, referring to damage caused by the internet.

    Minister reply

    Agreed with John Hayes about balancing risks and opportunities; referred to Online Safety Act 2023 covering illegal harms in various spheres. Emphasised importance of supporting start-ups and innovative companies while being conscious of risks.

  • Kirsty Blackman SNP Aberdeen North

    Question

    Inquired about the ownership of data gathered by AI, responsibility for ensuring its safety, copyright and intellectual property protections, impacts on climate change, and gender-specific harms. Asked if regulations can be brought forward without a long delay.

    Minister reply

    Acknowledged concerns about data ownership and safety, emphasised the importance of international collaboration in regulating AI. Mentioned the role of AI in predicting climate change but also its environmental impact due to electricity consumption. Promised to consider copyright issues and build trust in AI through the White Paper response.

  • Matthew Hancock Con

    Question

    Praised the Government’s approach on AI, highlighting its exponential growth and global impact, and asked for continued focus on empowering development of AI and setting guardrails.

    Minister reply

    Expressed gratitude to Matthew Hancock for past collaboration. Emphasised the UK's strong tech economy with over £2 billion in recent investment and 1.9 million jobs created since pre-pandemic levels.

  • Jim Shannon DUP Strangford

    Question

    Welcomed the focus on Bletchley declaration but expressed concern about censorship of unpopular positions and prioritisation of cyber-security.

    Minister reply

    Acknowledged AI's role in cybersecurity, mentioned recent launch of codes of practice for developers in the field. Highlighted importance of having regulators define and address industry-specific issues.

  • Steve Brine Con

    Question

    Asked about regulatory landscape for AI in healthcare, citing its potential support role but not replacement for clinicians.

    Minister reply

    Agreed on the importance of regulatory environment supporting innovation. Referenced visit to Aival and DeepMind's work in developing AlphaFold.

  • Mark Logan Con

    Question

    Asked about China’s role in AI safety commitments, referring to Tsinghua University.

    Minister reply

    Emphasised the importance of international collaboration including China. Mentioned that 16 companies globally signed a commitment for not deploying or developing models that test thresholds at the AI action summit.

Shadow Comment

Chris Bryant

The shadow Minister welcomed the economic opportunities of AI but criticised the Government's approach as lacking in regulatory measures. He questioned how shared risk thresholds would be drawn up and their legal force in the UK without legislation. He also raised concerns about companies marking their own homework regarding risks, state actors being 'malicious', integration with cyber-security strategies, and why no creative industries representatives were present at the summit despite its importance to a £127 billion industry. The collapse of the voluntary code of practice working group was highlighted as evidence of the Government's failure to follow through on commitments made six months ago. Bryant called for binding regulation of powerful AI companies requiring them to report before training models over capability thresholds and conduct safety testing, echoing advice from Government advisers.

Get updates like this by email
▸ Assessment & feedback
Summary accuracy