Project launch: Australian AI legal safeguards
Ah, marvellous. The University of Western Australia has launched "Autonomy under Law: Legal Guardrails for AI," a project title so perfectly academic it could double as the theme of a conference where nobody has actually been harmed by AI, only inconvenienced by a chatbot hallucinating a recipe. The project's stated purpose is to build "empirical evidence" on how "Australian-built techno-legal infrastructure" can "reduce the threat of AI harms." Note the word "infrastructure" — because a simple law would be far too déclassé for the twenty-first century. We shall build guardrails the way other nations build bridges: after the collapse, with concrete evidence of why the collapse was unavoidable.
As with all great Australian policy initiatives, the announcement contains no funding figures, no named investigators, no timeline, and no actual legal text. Just a university, a project name, and a vague but confident commitment to "empirical evidence" — which in this context appears to mean "we will count the harms while they happen, then publish a PDF." The phrase "Australian-built" is doing heroic work, presumably to distinguish these guardrails from imported European or American guardrails, as if AI harm were somehow sensitive to national borders. One can already imagine the committee meetings: "Should the guardrail be a code of conduct or a statutory obligation?" "Let's commission another project to find out."
The dark comedy is that the "threat of AI harms" is itself treated as a hypothesis requiring evidence, rather than the current state of affairs. Perhaps the researchers will eventually produce empirical confirmation that by the time you have gathered enough evidence to justify a guardrail, the AI has already moved, replicated, or been given its own legal personality. Still, credit where it's due: the university has bravely named its initiative "Autonomy under Law," as if the machines had politely agreed to submit to legal process. Somewhere, an AI is laughing — but without empirical evidence, we cannot confirm it.