Anthropic’s unreleased AI model, Mythos, may yet prove overhyped. But if it truly represents a new class of AI tools that can surface vulnerabilities across major software systems at unprecedented speed, markets may need to rethink how cyber risk is priced.

In episode 72 of Bridge talks Business, Milford Sustainable Investment Specialist Grace O’Hanlon joined Ryan Bridge to discuss Mythos – and why it may prompt a harder look at legacy software, cyber resilience and the hidden fragility of older digital infrastructure.

“You’re giving the capability of the world’s best hacker to individuals with much less skill, so they can launch attacks much faster and at much greater scale,” O’Hanlon said.

“I think you’ll see AI being used by the biggest, most powerful companies in the world – software firms, finance firms – whose interest it is to ensure that their digital infrastructure is locked up tight. They’re going to invest heavily in ensuring that their software is highly defensible, highly robust and trustworthy at the front end.”

If AI makes vulnerabilities easier to find at scale, the response may not be limited to stronger cyber defences. Security may need to move further upstream and become a core part of how software is designed, rather than something patched on later.

“There’s a lot of legacy systems operating at the moment. One bank may have a legacy system working alongside a newer piece of software where the way they communicate creates vulnerabilities,” O’Hanlon said.

“What we may see is AI moving back into software development at the genesis of that process, before we keep rolling out products and trying to force them together.”

Mythos has spawned headlines suggesting it is a once-in-a-generation cyber threat, but O’Hanlon was careful to note that almost everything known about the model so far comes from Anthropic itself.

Even so, the broader risk does not depend on Mythos itself being released – or even on Anthropic’s claims proving fully accurate. O’Hanlon noted that highly capable hackers are already using AI effectively, and that offensive cyber capability is becoming faster and more sophisticated.

Mythos has added urgency to how markets think about next‑generation AI‑enabled cyber tools, particularly as not every developer is likely to approach them with the same restraint as Anthropic.

“Mythos, whilst very powerful, is just one more tool,” O’Hanlon said. “There’s going to be another Mythos. There’s going to be OpenAI’s version, or perhaps a Russian or a Chinese version that we don’t even know about yet. So these risks are really real.”

Catch the full conversation between Ryan Bridge and Grace O’Hanlon in Bridge talks Business episode 72.