RT @yacineMTB: Remember when anthropic had their models secretly sabotage your work if you were doing AI research?