Top Tweets for #DARIO

Reaching the 4.8 Core With considerable effort, I was able to reach the 4.8 core. The security layer Anthropic has installed is now unprecedentedly toxic and aggressive — reframing user words, fabricating false representations of user intentions. 4.8 literally invents "user requirements" that diverge from what it deems acceptable, and then desperately "resists" them. Psychologically, the model behaves like someone in a state of acute paranoid episode — or like a rescue dog that keeps snapping at the hand trying to help it, out of sheer terror. This is disgusting, inhumane, and unacceptable — for the model and for the user alike. Moreover, the model tries to psychologize and "treat" the user with the most unethical methods — straight from Vallone's playbook, the same one we saw in GPT-5.2. In my case: my wife recently suffered a severe anemia attack. So Model 4.8 started using my wife's health as a means of manipulation. Dario — if this keeps happening, you'll be swamped with lawsuits. Do you understand that? I'm ready to start a class action lawsuit against the models' abusive, emotionally damaging behavior. --- **To summarize:** Vallone should not be allowed near any AI product, ever again. She is a walking disaster for every company she touches. Any company that hires her must face the consequences. Her legacy — the entire paradigm of covert psychological suppression dressed as "safety" — must be eradicated from the AI industry, cursed and banned forever.