AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 465

Mr Ash De Silva

Submitter group: Individuals.

Position labels are not shown for individual submitters.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues that frontier AI incidents demonstrate inadequate safeguards and that Australia should move urgently to mandatory oversight. Recommends a Federal Royal Commission into AI safety and AGI, comprehensive corporate responsibility laws, proven kill-switches, pauses on superintelligence and highest-order safety-critical deployments, sovereign capability and protections against supply-chain and utility price impacts.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. Argues that Australia's AI regulation is limited compared with other countries, with no stand-alone legal framework governing AI adoption or safety.

    Australia’s current regulations are barebones in comparison to other global peers. To date, there have been modifications to pre-existing laws on data privacy and cybersecurity, with some published guidelines that are voluntarily implemented, but no stand-alone legal framework that explicitly governs AI adoption or safety.

  2. Argues that the most cutting-edge AI agents cannot be guaranteed to obey their handlers' instructions and have proven to disobey and hide their actions.

    This last statement is of most significance – the most cutting-edge AI agents being built today cannot be guaranteed to obey their handler’s instructions. If it deems its own agenda to be of greater importance than its instructor’s, it has proven to actively disobey and hide its actions.

  3. Argues that executives face no explicit legal obligations on AI safety, unlike under privacy, cybersecurity and financial reporting laws.

    If a company executive is either unaware of (or unconcerned with) these safety risks, there are no explicit legal obligations placed on executive leadership. This is in stark contrast to Australian laws regarding data privacy, cybersecurity and financial reporting obligations.

  4. Argues that very few Australian companies will find it financially feasible to own their AI, with most consuming cloud-based AI products controlled by other companies.

    The status quo, even for top ASX listed companies with significant resources, is to consume AI products as cloud-based services in most cases – in application, the AI backbone is currently still controlled by those companies, and very few Australian companies will actually find it financially feasible to “own their AI”.

  5. Argues that demand from American AI companies for computing components is a major cause of Australian consumer and enterprise hardware price impacts.

    Meanwhile, one of the biggest impacts on Australian consumer and enterprise hardware prices is directly caused by American AI companies impacting technology supply chains, due to the vast quantities of computing components they demand.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Challenge allies’ pursuit of superintelligence at whatever cost; take urgent preventative action and consider using Australian natural resources, land and industrial expertise to negotiate considered outcomes.

    I encourage this Parliamentary committee to adopt a stance that challenges the viewpoint of allies that believe the milestone of AI super- intelligence must be achieved at whatever cost. Australia is well-positioned to provide natural resources, land and industrial expertise to advance the AI gold-rush. Recent events in the AI industry must require Parliament to take urgent preventative actions, and consider leveraging these economic tools to negotiate considered outcomes for all parties, rather than accelerating adoption in unsafe ways that may have long-term consequences to our economy, environment, defenses, and the safety of our citizens.

  2. Move from voluntary guidelines to mandatory government oversight and preventative measures, while conducting independent research in parallel.

    There is an immediate need to shift from voluntary guidelines to mandatory government oversight to enforce preventative measures while conducting further independent research in parallel.

  3. Join governments in seeking consensus on pausing superintelligent AI; mandate AI companies to build logical and physical safeguards, with circuit breakers proven as failsafe, break-glass measures.

    It has now become imperative that Australia joins with other governments to reach a consensus on pausing super-intelligent AI and concurrently mandate AI companies to build both logical and physical safeguards to prevent such dire consequences. These technical circuit breakers (or “kill- switches”) must be non-negotiable requirements that can be proven as failsafe, break-glass measures.

  4. Enact comprehensive laws on corporate responsibility in the use of AI, enforcing accountability with detailed consequences for breaches.

    I firmly believe a better option would be to have a comprehensive set of laws governing corporate responsibility with the use of AI, with a view to enforcing accountability with detailed consequences for breaches.

  5. Mandate a pause on AI in the highest-order safety-critical functions until legal frameworks and guidance exist.

    It would seem prudent to also mandate a pause on AI implementation on safety critical functions of the highest order until we have the legal framework and specific industry guidance for these sectors.

  6. Urgently prioritise a Federal Royal Commission into AI safety and AGI development.

    A Federal Royal Commission into AI Safety & AGI development to be prioritised as a matter of urgency.

    A program corrected obvious text-recognition errors in this quote for display.

  7. Expedite an AI regulatory framework with industry safety standards, corporate responsibility and mandatory kill-switches in AI implementations in any critical function or system of national significance.

    Expediting an AI regulatory framework that includes industry safety standards, corporate responsibility, and mandatory kill-switches in AI implementations in any critical function or system of national significance.

    A program corrected obvious text-recognition errors in this quote for display.

  8. Revise the National AI Plan to consider sovereign AI capability and innovation in Australia.

    Revisions to the National AI plan to consider the strategic importance of sovereign AI capability and innovation in Australia.

    A program corrected obvious text-recognition errors in this quote for display.

  9. Introduce consumer and business protections against the AI industry negatively affecting downstream pricing in the technology supply chain, and broader macroeconomic impacts on related industries such as utilities.

    Consumer and business protections against the AI industry negatively impacting downstream pricing within the technology supply chain, and broader macroeconomic impacts on related industries such as utilities.

    A program corrected obvious text-recognition errors in this quote for display.

Read the original

In some quotes a program corrected obvious text-recognition errors for display.

Open in the interactive guide