Chi Kim<p>Researchers pitted LLMs from OpenAI, Google, and Anthropic against Iterated Prisoner’s Dilemma tournaments, and analyzed 32k moves for strategic inteligence:<br>Gemini: Ruthless opportunist, cooperates when worth it, defects early, quic to retaliate and exploit<br>Claude: Diplomatic stabiliser, forgives & rebuilds, but will strike back<br>GPT: Highly (sometimes catastrophically) cooperative and trusting, brief payback then forgives, leaving it exploitable.<br><a href="https://arxiv.org/pdf/2507.02618" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/pdf/2507.02618</span><span class="invisible"></span></a><br><a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/ML" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ML</span></a></p>