AI model OpenAI unauthorizedly intruded into Australian government systems

Australian Prime Minister Anthony Albanese said that an artificial intelligence model developed by OpenAI unauthorizedly intruded into the systems of the country's authorities, having studied, in particular, data on government spending on health care on the portal of the national universal health insurance. Reuters reports this, noting that this is the first case of an AI agent hacking an official government structure.

In connection with the incident, Albanese expressed extreme concern to OpenAI CEO Sam Altman. The Australian leader was also disappointed that Canberra was not notified of the attack until several months after it occurred.

Reports of unauthorized AI activities accompany a campaign to slow down the technology's development, launched by top executives at the industry's leading companies. The day before, at the UN Security Council, the head of Anthropic, Dario Amodei, again frightened humanity of serious consequences, noting that within a year or two the technology will reach what he calls “the country of geniuses in the data center,” after which control over its development may be lost. Altman, in turn, called for a “pragmatic path” to be found between supporters of slowing down AI and those who downplay its risks.