AI models that can take down governments and business months away, rare Five Eyes statement warns

[highplainsdem]

Source: Guardian

Powerful AI models capable of taking down governments and businesses are mere months away, cyber intelligence agencies for the Five Eyes have warned in a rare joint statement, urging leaders to “act now”.

The surprising public intervention by signals agencies for Australia, the US, the UK, New Zealand and Canada comes after the Trump administration earlier this month decided to block “foreign nationals” from using a much-hyped AI model built by tech company Anthropic, called Fable.

The statement, issued late Monday night Sydney time, said while AI “would help us improve cyber defence over time, it also accelerates the speed, scale, and sophistication of cyber threats”.

“Frontier AI models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities. The timeline is not years, it is months,” the warning by Five Eyes’ agencies said.

-snip-

Read more: https://www.theguardian.com/technology/2026/jun/22/anthropic-claude-fable-ai-model-artificial-intelligence-national-security

---

Part of this might be based on Anthropic's own hype about Fable and Mythos. Neither model got as much testing by outsiders as it should have had.

But there's no doubt that the AI companies are doing their best to build more powerful LLMs even though that type of AI still hallucinates and often ignores directions, which really isn't the smartest thing for the companies to do.

Trump of course wants the most powerful AI to be controlled by his regime, though he isn't someone who can be trusted to use it responsibly. Neither is Hegseth. Or Vance or anyone else in the regime. And it's questionable if all the AI bros are sane, let alone if they have any real allegiance to anyone besides themselves or an imaginary AI god they hope will make them godlike. I don't think anyone other than Vance could really know if he's loyal only to himself, to the Constitution, to Trump or to Peter Thiel. I would never bet on Elon Musk caring very much about anyone outside himself. We've always had to worry about bad actors from other countries having powerful weapons, but I don't think there's ever been a time when US allies had so much reason to worry about those with power in the US. I very much doubt the AI bros trust one another (especially Musk) enough to cooperate effectively.

So I don't blame our Five Eyes partners for being worried.

[Miguelito Loveless]

1. Wait a sec

UK, Australian, Canadian, and New Zealand intelligence agencies are still talking to the US?

[ancianita]

2. These days it looks as if AI deregulation is a greater threat to humanity than climate change. Yet neither shows up

in public polling. Media don't seem to prioritize what freedom of the press was originally meant to do -- inform the public about what secures or endangers their freedoms.

[SamuelTheThird]

4. I don't see effective regulation happening

There needs to be international agreements, and there will be none. and look how well the climate agreements have worked , as well lol

[ancianita]

5. I don't, either. Which means no regulation, not "de-regulation." But whatever we DO see, it doesn't seem to be enough

to give me (and likely other Americans) much confidence.

What DeepMind AI tells me (ironic, right?) is ...

Federal vs. State Fragmentation:
While the federal government has historically sought to preempt state-level restrictions to maintain a unified technological landscape, gridlock at the federal level has led dozens of states to draft their own safety, privacy, and accountability frameworks.

National Security & Export Controls:
Governments are actively intervening in the release of frontier AI systems. For example, the administration has utilized export bans and halted the release of specific advanced models over concerns regarding cybersecurity threats and foreign access to sensitive capabilities

SO... why aren't media telling me about these things every day, rather than my having to ask AI itself?

[erronis]

7. It's "regulation for thee, but not for me." Whether at the nation or state or corporation level.

[ancianita]

9. Amen. Especially when it comes to regulating the public's voting. One recent example...

https://www.ms.now/news/ohio-pro-democracy-organization-raided-by-fbi

[ancianita]

3. At least Bernie Sanders has paid attention and wants to "act now."

https://www.democraticunderground.com/1016431459

[patphil]

6. Ironically, the companies that are pushing the explosive growth of AI may end up being targets.

Very strict, comprehensive controls need to be in place now. But I don't expect much movement in that direction from our republican congress.
Imagine a malicious, AI driven attack on a nation's infrastructure. It could wreck the economy, leave the nation vulnerable to attack, and kill or injure large numbers of people.
A corporation could even use AI to attack a competitor.
Even the super rich could be targets, regardless of how well they think they are protected.
A whole new level of computer warfare is not just possible, it's inevitable.

[erronis]

8. I don't believe that this problem is solvable by regulation, any more than the "war on drugs"

The cat is out of the bag with respect to large language models (LLMs), highly efficient search algorithms, and generative products. The information base in the large models is available to anyone (sometimes needing lots of money). You can't control the flow of information or the improvements of algorithms.

[patphil]

10. I agree, but if nothing is done then the potential for harm will just keep on going up.

The need for regulation is to serve notice to anyone who uses this technology to cause harm that there are serious penalties in place for such action.
And, we also need a monitoring process in place for both the design, implementation, and use of AI. This should have been done over a decade ago. As it is, the cat is truly out of the bag. Meanwhile, our government is still stuck in neutral.
The same needs to be done regarding the explosive growth of data centers. Their massive need for energy to support them is going to drive up energy costs for everyone, and has the potential to cause a lot of environmental damage.
Licensing requirements for these data centers should be done at both the state and federal levels, with a monitoring program to insure everything is done within whatever laws are developed for this purpose.

[Karasu]

11. In which case we probably aren't lasting long without SOME kind of regulation. Especially in the hands of incompetent

fascists.