IN NEWS: @promptfoo raises a $18.4M Series A led by @insightpartners. We spoke with @iwebst (Founder & CEO) on AI safety and adversarial testing: "We've trained models that behave as adversarial users targeting specific risk areas like PI leaks or toxicity, or talk about competitors." "What this amounts to is we wind up having thousands and thousands of conversations where the attacker is just trying to wiggle its way around some of the guardrails."