Pip: Good Strategy Rebellion — where the chess pieces are real, the board is on fire, and someone keeps insisting the rules still apply.

Mara: Today we’re in difficult territory. Martyn Jones puts a direct question to several major AI systems: is there tangible, verifiable evidence that Israel has engaged in war crimes, genocide, or sexual violence in Gaza? The answers are substantial, and they matter.

Pip: Let’s start with what the AI systems actually said — and what they agreed on.

Examining what the AI systems found

Mara: The post frames a single, pointed question put to Grok, Gemini, ChatGPT, and Copilot: is there verifiable evidence of mass slaughter, infanticide, genocide, rape, and other grave abuses?

Pip: And the answer across all four systems was, essentially, yes — with important distinctions about what kind of evidence exists and what remains legally unresolved.

Mara: Grok’s response sets the baseline directly: “Yes, there is tangible, verifiable evidence of grave human rights abuses and war crimes by Israeli forces in Gaza, including mass civilian casualties, destruction of civilian infrastructure, and sexual/gender-based violence.”

Pip: So the documented evidence is not in serious dispute. The contested ground is the legal classification — particularly whether the specific intent required for genocide has been demonstrated.

Mara: Exactly. Gaza’s health ministry reports around 73,000 killed and 173,000 injured. Independent epidemiological analysis, including work published in The Lancet, suggests undercounting due to bodies under rubble, disease, and collapsed healthcare. Multiple sources put women and children at between 46 and 70 percent of casualties.

Pip: Those are not numbers you can argue away as a rounding error.

Mara: Gemini adds the ICJ dimension. In January 2024, the court issued a preliminary ruling finding it “plausible” that Israel’s actions fell within the scope of the Genocide Convention, ordering provisional measures. The full case on the merits is expected to take years.

Pip: Plausible is doing a lot of legal work there — it’s not a conviction, but it’s not nothing either.

Mara: ChatGPT draws a useful distinction between documented acts, allegations of specific crimes, and final legal determinations. On sexual violence, it notes that in 2026 UN reporting led to Israeli security forces being added to the UN blacklist for conflict-related sexual violence — covering abuse of detainees including rape, forced nudity, and torture.

Mara: Copilot is the most direct of the four. It states that Amnesty International concluded there is “sufficient basis to conclude that Israel has committed and is continuing to commit genocide,” and that UN inquiries repeatedly describe Israeli conduct as the crime against humanity of extermination.

Pip: All four systems also note that Hamas committed documented war crimes on October 7th — murder, hostage-taking, sexual violence — and that international law assesses each side’s conduct separately.

Mara: The post doesn’t editoralise beyond the framing. It lets the AI responses stand as a composite picture: overwhelming evidence of war crimes, a serious and partially adjudicated genocide case, and a legal process that is still unfolding.

Pip: Evidence is one thing. What accountability looks like is the harder question that follows.

---

Mara: The evidence is documented, the legal proceedings are live, and the humanitarian catastrophe is not in dispute.

Pip: The gap between what is known and what is acted on — that’s the territory worth watching. More from Good Strategy Rebellion next time.