We do model security + behavior research, especially for cyber. Abliterating, RL, SFT etc are parts of studying what models actually do for cyber tasks.
Replying to @OrcaRouter and @Stoxx_Beaver
Are you sure it's not Sam Altman behind this account? For somebody claims to "uncensor" you spend an inappropriate amount of time yapping about safety and praising its lobotomite results.
