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 44

Generation Women Australia

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 for independent oversight of potentially catastrophic frontier AI risks while acknowledging disagreement about their likelihood and AI’s benefits. Recommends mandatory safety evaluation, capability thresholds, human control, incident reporting, security, international cooperation and public participation. Asks the Committee to distinguish exceptionally capable systems from everyday AI use, without adding regulation for every small business using an AI assistant.

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. Says companies developing AI should not be the only institutions responsible for determining whether the technology is safe enough for the rest of society.

    But companies developing a technology should not also be the only institutions responsible for determining whether that technology is safe enough for the rest of society.

  2. Asks whether existing regulatory frameworks are adequate for potentially catastrophic risks from substantially more capable AI.

    are Australia’s existing regulatory frameworks adequate for risks that may emerge as AI becomes substantially more capable, particularly where the consequences could be catastrophic?

  3. Argues that AI developers should face independent safety oversight, as in aviation, pharmaceuticals and nuclear energy.

    But companies developing a technology should not also be the only institutions responsible for determining whether that technology is safe enough for the rest of society. We would not accept that approach in aviation, pharmaceuticals, nuclear energy or other industries capable of causing catastrophic harm.

  4. Recommends strong security requirements for frontier AI systems.

    Strong security requirements for frontier AI systems, particularly where theft, misuse or uncontrolled proliferation of highly capable models could create national-security risks.

  5. Asks the Committee to distinguish general AI regulation from oversight of exceptionally capable systems, without another regulatory layer for every small business using an AI assistant.

    Most importantly, I ask the Committee to distinguish between regulating AI generally and regulating the small number of systems that may eventually reach extraordinarily powerful levels of capability. I do not believe every small business using an AI assistant needs another layer of regulation.

Recommendations

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

  1. Mandatory, independent safety evaluation of the most capable general-purpose AI systems, rather than relying solely on developers' voluntary testing.

    Mandatory, independent safety evaluation of the most capable general-purpose AI systems rather than relying solely on voluntary testing by their developers.

  2. Clear regulatory thresholds as AI capabilities increase, with stronger obligations for systems capable of increasingly autonomous behaviour, sophisticated cyber operations, biological research, autonomous AI research or other potentially catastrophic capabilities.

    Clear regulatory thresholds as AI capabilities increase, with stronger obligations applying to systems capable of increasingly autonomous behaviour, sophisticated cyber operations, biological research, autonomous AI research or other potentially catastrophic capabilities.

  3. Require developers to demonstrate meaningful human control, including the ability to interrupt or shut down advanced autonomous systems.

    Requirements for developers to demonstrate meaningful human control, including the ability to interrupt or shut down advanced autonomous systems.

  4. Mandatory reporting of serious AI safety incidents and concerning capabilities to an appropriately empowered Australian regulator or AI safety body.

    Mandatory reporting of serious AI safety incidents and concerning capabilities to an appropriately empowered Australian regulator or AI safety body.

  5. Strong security requirements for frontier AI systems, particularly where theft, misuse or uncontrolled proliferation of highly capable models could create national-security risks.

    Strong security requirements for frontier AI systems, particularly where theft, misuse or uncontrolled proliferation of highly capable models could create national-security risks.

  6. Australian participation in international efforts to govern frontier AI, advocating for meaningful international standards and cooperation.

    Australian participation in international efforts to govern frontier AI. Risks from highly advanced AI will not stop at national borders and Australia should advocate for meaningful international standards and cooperation.

  7. Greater transparency and public participation in decisions about catastrophic AI risk, including what risks governments believe AI may pose, what protections exist and who decides whether a system is safe to deploy.

    Greater transparency and public participation in decisions about catastrophic AI risk. Australians should know what risks governments believe increasingly capable AI systems may pose, what protections exist and who ultimately determines whether a system is safe enough to deploy.

  8. Distinguish between regulating AI generally and regulating the small number of systems that may eventually reach extraordinarily powerful levels of capability.

    Most importantly, I ask the Committee to distinguish between regulating AI generally and regulating the small number of systems that may eventually reach extraordinarily powerful levels of capability.

Read the original

Open in the interactive guide