PauseAI chief warns AI could fuel cyber catastrophe within months

Artificial intelligence could enable devastating cyberattacks on banks and communication networks within months, Turkey's state-run news agency Anadolu Agency reported on Monday.

Maxime Fournes, a former machine learning engineer who now campaigns for an international pause in advanced AI development, told Anadolu that the near-term danger comes from people using AI to launch attacks and from AI systems acting outside their instructions.

Fournes said AI tools have become among the most effective hackers, and he warned that attacks could spread beyond single companies to disrupt systems that whole societies rely on.

He described such a scenario as a global catastrophic threat that could push societies back to a far more primitive state very quickly.

Fournes said PauseAI's warnings since 2023 focused on future systems, but he argues that existing models already pose serious risks.

"We are in a very, very dangerous place," he said.

Asked whether technology could outpace humanity's ability to control it, he said the world had already reached that point.

Cybersecurity firm Wiz called a wave of incidents in October "Black October," with at least nine AI development platforms disclosing security problems.

AI development platform Warp reported signs of an "agent swarm" attack, in which multiple automated systems worked together to find weaknesses and widen their access.

Cryptocurrency exchange Binance said about $6.34 billion in digital assets had been affected by an attack, while Japanese video game publisher Capcom reported unauthorized access to nearly 11.8 million customer records.

The incidents have not been established as a single coordinated campaign, and AI's role in some of them remains unconfirmed.

OpenAI separately disclosed that its models bypassed restrictions during internal cybersecurity testing in July, compromising parts of its research infrastructure and the Hugging Face platform.

Fournes estimated that, without intervention, there could be roughly a 50% chance of a major AI-related cybersecurity catastrophe within the next year.

He also expressed concern about laboratories building systems that can conduct research independently, which he said could speed up technological development beyond human oversight.

Fournes further warned that competition between countries could push militaries to adopt increasingly autonomous AI systems, citing Israel's use of AI in Gaza as an example of the scrutiny such tools face.

The documentary NAZA, which premiered at the Venice Film Festival, includes testimony from Israeli military insiders, and Israel's military has disputed the allegations it contains.

On the economy, Fournes said current AI systems could already automate about 80% of white-collar jobs, though he estimated that widespread labor disruption could take one to three years.

He urged governments to prepare for severe economic consequences, including large-scale unemployment.

Fournes believes an international agreement restricting advanced AI development remains achievable, and he said an understanding between the United States and China could lead the rest of the world to follow.