Exclusive: Anthropic is testing ‘Mythos,’ its ‘most powerful AI model ever

Do you think that Anthropic should gatekeep this model ? I mean it’s their product (now that they violated so much intellectual property rights but anyway) so it’s their right but is it in favour of the common good ? Will Anthropic even succeed in serving Mythos only to good actors ? Or will bad actors manage to jailbreak Mythos rules, or gain legitimate access? And what definition of good actors does Anthropic have? I think that there is a risk of Anthropic doing favoritism with who they chose to give access to Mythos (for instance giving access to Mythos to Visa but not Mastercard because of other pressures or interests). If Anthropic was to open-source it, it would surely be used by hackers, but I see it as a potential race between hackers and cybersecurity experts. My take is that now that this model exists, it should be released to everybody, to improve the cybersecurity strength of digital services. Especially to small players that do not have a lot of resources to allow to cybersecurity.