A new AI model from Anthropic called Mythos sparked concerns over vulnerability detection capabilities, though cybersecurity experts say the same outcomes can be replicated using AI systems already in public use.