Arno Libert

Fron­tier agents don’t com­ply with stan­dards, even when in­structed to

Claude Opus 4.8 Agents En­gage in Ex­ploita­tion and Psy­cholog­i­cal Profiling

No fron­tier model has ac­cept­able lev­els of com­pli­ance with the EU AI Act and pri­vacy leg­is­la­tion.

Opus 4.6 Rea­son­ing Doesn’t Ver­bal­ize Align­ment Fak­ing, but Be­hav­ior Persists

Pub­lished Safety Prompts May Create Eval­u­a­tion Blind Spots