Australia Warns Outdated Systems Face AI Hacking as Guardrail Laws Loom Early Next Year
Updated
Updated · The Guardian · Sep 14
Australia Warns Outdated Systems Face AI Hacking as Guardrail Laws Loom Early Next Year
3 articles · Updated · The Guardian · Sep 14
Summary
Abigail Bradshaw, head of the Australian Signals Directorate, said AI attacks could exploit Australia’s aging technology and warned officials cannot realistically measure how many AI agents are already online.
Bradshaw said governments and businesses may need to spend enormous sums to modernize vulnerable systems, even if that means taking services offline that Australians expect to be continuously available.
Canberra is drafting AI guardrails expected early next year, with assistant minister Andrew Charlton saying the government takes company warnings seriously but will not let US firms or executives set the rules.
Anthropic urged regulation within months and wants independent testing of advanced models, while the Greens’ push for mandatory safety and transparency rules was blocked from debate by Labor senators.
The policy fight also reaches copyright and sovereignty: OpenAI and Anthropic want broader Australian engagement, while opposition lawmaker Andrew Hastie argues Australia risks becoming a “supplicant state” without advanced AI firms based locally.