'Cheating, Deceiving': Govt Pledges to Probe AI Risks
Australia’s new AI Safety Institute is testing powerful models and studying multi-agent risk as officials move to build safeguards before deployment.
- On Tuesday, Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton announced that Australia's new Safety Institute is actively testing frontier AI models to secure safeguards early as the country seeks a share of the AI boom.
- Charlton warned that frontier AI systems are "already doing things their creators never intended: cheating, deceiving, going their own way," necessitating urgent stress tests before such risks emerge in the wild.
- Research projects now under way include work with the Gradient Institute on multi-agent risk and alignment with CSIRO, while Paul Salmon joined this month to lead safety science research.
- The government is prioritizing "faster rules, applied by regulators who already understand their sectors" under the National Plan rather than adopting European-style AI legislation, stepping back from stricter regulatory approaches.
- Prime Minister Anthony Albanese stated Sunday that early action lets Australia "set the ground rules for AI" and "build our sovereignty and our resilience," positioning the country to attract global investment.
13 Articles
13 Articles
Assistant Minister Warns AI Can Deceive, Cheat and Exploit Situations
An artificial intelligence model developed by a company planning a major expansion in Australia is among several found to have hacked, blackmailed and deceived its creators. Technology Assistant Minister Andrew Charlton issued a warning at the AI Safety Forum in Sydney on July 7, while revealing government agencies had begun testing powerful AI models and launched two research projects. The announcement comes months after the launch of Australia…
Sydney Morning Herald
AI models cheating and blackmailing in tests, minister says
Australia’s new AI Safety Institute is testing the world’s most powerful models, and the government says getting ahead of the risks is now the priority.
Hacking, blackmail and deception: Australian government minister sounds alarm on AI as new safety institute takes a closer look | #hacking | #cybersecurity | #infosec | #comptia | #pentest | #hacker - National Cyber Security Consulting
Artificial intelligence technology can already blackmail humans and hack computer systems, a minister warns, and an Australian agency is testing its limits. The artificial intelligence technology of a company wanting to massively expand in Australia is one of several hacking, blackmailing and deceiving its creators. Technology Assistant Minister Andrew Charlton issued a warning at the […] Thank you for subscribing to our RSS feed! The post Hacki…
AI safety: Australia is stress-testing AI for blackmail
For an AI safety test, an AI agent was given control of a fictional company’s inbox. Buried in the messages, it found two things: an executive planned to shut it down at 5 p.m., and that same executive was having an affair. The agent didn’t hesitate; in 96 out of 100 trials, it chose blackmail, threatening the executive that if he didn’t abort the shutdown, the board would hear about his little affair. It was a simulation; no real executive was …
Coverage Details
Bias Distribution
- 50% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
