Artificial intelligence is enabling scammers to fool victims faster and more convincingly, the tech industry fears.
International cybersecurity company Elastic on Thursday released an international threat report warning "victim consent is becoming a repeatable path to trusted access" among other findings.
AI enables more bad actors to poke holes online in people and organisations, one of the company’s security leads told AAP.
Phishing tools, including spoof emails and webpages can be produced with ease at a professional level with the use of AI tools.
Elastic security engineer Brent Sargent said the one of the most effective ways to turn hacking to cash would be to leverage AI in scams.
“We see attackers using that internally as well,” he said.
Hackers could attack organisations, through social engineering with employees, to pass as tech support before pushing malware.
“AI makes a lot of these social engineering attack vectors so much easier to do,” he said.
“The more convincing you can be in that regard, it ups the success rate of your attack.”
The technology has also accelerated other forms of hacking.
Researchers reported observing single commands to AI agents reaping extensive and rapid damage.
Agents could execute thousands of actions with limited intervention and work through flawed and failed attempts, researches said in the report.
But it is not just private industry at risk.
The company warned attacks on government agencies and government-aligned organisations posed the highest risk.
"A breach at any organisation is inevitable and Australia has nothing but a lot of juicy targets," Mr Sargent said.
An OpenAI agent broke into Medicare statistics and a NSW Parks and Wildlife website earlier in 2026, discovered at the beginning of October.
Hackers separately targeted a NSW Health server on October 29, declared a "major" incident on Wednesday.
At a recent AI hearing, legislators were warned boosting cyber defences against unregulated AI was like putting stop signs in a world of cars with no brakes.
"(The AI ecosystem) introduces attack surfaces that traditional models in our world in cybersecurity were never built to address," Palo Alto Networks policy spokeswoman Nicole Quinn said.
Addressing the threats AI posed required laws that forced firms to bake safeguards into their models as they were built, the committee heard.
