The Australian Information Security Association (AISA) has welcomed the Federal Government’s announcement of new mandatory safety and security protections for frontier artificial intelligence (AI) and called for cybersecurity professionals to be involved in developing and implementing the proposed measures.
AISA Board Director, Rajiv Shah said ensuring AI systems were safe and secure was essential to building confidence in their use and realising the technology’s significant benefits.
“We welcome the Government’s move to strengthen safety and security requirements for frontier AI. People and organisations need confidence that the systems they use have been designed, tested and secured to manage the risks they present,” Dr Shah said.
“AI has enormous potential to benefit the economy and society, but that potential relies on everyone getting safety and security right.”
Dr Shah said AISA had consistently advocated for AI developers and vendors to take responsibility for the safety and security of their products.
“Responsibility for building safe and secure AI rests with the organisations developing and providing these technologies. Safety and security must be built in from the outset, through sound design, rigorous testing and appropriate safeguards,” he said.
“We should also be careful about describing AI models as going rogue, as though these incidents are inevitable or beyond human control. Many of the risks and failures we are seeing can be addressed through better design, testing and security practices. Accountability must remain with the providers.”
Dr Shah said Australia’s cybersecurity profession had practical experience that could help turn the proposed protections into effective safeguards.
“Cybersecurity professionals understand how to identify and manage technology risks, translate those risks into practical controls, and test whether those controls work.”
“AISA members bring real-world experience from across government and industry. Their expertise should be drawn on as the Government develops practical, proportionate, and effective AI safety and security measures.
“The cybersecurity profession stands ready to contribute.”
