> The surge in AI-generated reports comes as Anthropic last month launched Mythos, its new cyber AI model, which it says can find software flaws faster than humans.
Is that true? All this started last month and only when after Mythos?
Also kind of interesting they mentioned Mythos by name twice but no other models that do the same thing like OpenAI's Daybreak.
From another article:
> While Anthropic’s model is really good at finding software vulnerabilities, so are other models. The UK’s AI Security Institute found that OpenAI’s GPT-5.5, already generally available, is comparable in capability. The company Aisle reproduced Anthropic’s published results with smaller, cheaper models.
While this might not be Anthropic PR, it definitely reads like it or by a journalist who did not do their due diligence and thinks Mythos will literally hack their grandma's phone today.
altmanaltman•40m ago
Is that true? All this started last month and only when after Mythos?
Also kind of interesting they mentioned Mythos by name twice but no other models that do the same thing like OpenAI's Daybreak.
From another article:
> While Anthropic’s model is really good at finding software vulnerabilities, so are other models. The UK’s AI Security Institute found that OpenAI’s GPT-5.5, already generally available, is comparable in capability. The company Aisle reproduced Anthropic’s published results with smaller, cheaper models.
https://www.theguardian.com/commentisfree/2026/may/08/how-da...
While this might not be Anthropic PR, it definitely reads like it or by a journalist who did not do their due diligence and thinks Mythos will literally hack their grandma's phone today.