To clarify - CB2 is the "can make novel bioweapons" threshold! ☣️
Also NEW: Mythos-5 is significantly stronger than Mythos-preview.
For CB (chemical and biological risk) Anthropic says that Mythos 5 is "near the border" of their Responsible Scaling Policy and Frontier Compliance Framework
This is fascinating. This is what made Anthropic decide that it's near CB2 - Without AI tools the task they were given was estimated in 40-95 days, with Mythos 5, two person team accomplished this task in 16 hours. While being generalists!
To clarify - CB2 is the "can make novel bioweapons" threshold! ☣️
To clarify - CB2 is the "can make novel bioweapons" threshold! ☣️
Also NEW: Mythos-5 is significantly stronger than Mythos-preview.
For CB (chemical and biological risk) Anthropic says that Mythos 5 is "near the border" of their Responsible Scaling Policy and Frontier Compliance Framework