Satcove News
Original research
Studies run by Satcove: the same questions put to six AIs, with the method and the raw answers.

World
Six AIs, One Market Crash: They Agree on the Numbers, Not on the Causes
We gave six AIs the same tech-crash headline. Their figures match, their explanations clash, and their odds are nearly identical.
Science
Six AIs Put Low Odds on a Quantum Break of RSA or ECC by 2030, and Split on Everything Else
Will quantum computers beat classical ones by 2030, break Bitcoin, or change medicine? We gave six AIs the same dossier and seven dated questions, three runs each. They converge on low odds for a near-term cryptographic break, but that reflects a shared dossier. On the rest, model medians differ up to ninefold, and a single model can move by tens of points between its own runs.

World
Six AIs on Four Middle East Risks by End-2027: Gaza, Lebanon, Iran and Saudi-Israel, Three Runs Each
In April, six AIs gave a vague "42% agreement" on Middle East peace. We reran the test with six dated, checkable questions, three runs per model. All six say comprehensive peace is unlikely; on single items their medians differ by up to 37 points, and one model alone moves by up to 45 points between its own runs.

Economy
Ueno Bank Collapse in Paraguay: Why It Is a Local Crisis, Not a Global One
Paraguay's central bank stepped in at Ueno Bank on 3 November. What it means for 3 million customers, the guarani and markets, and why a global crisis is unlikely.

World
Nine People, 100 Days Adrift: What Six AIs Say About Their Chances of Survival
Five of nine relatives survived about 100 days adrift in a 20-foot skiff. We asked six AIs how likely that was, and whether AI could have found the boat.
Politics
Six AIs Saw the Democratic Majority Coming. Voters Added a Verdict on Trump
Six AIs gave Democrats the edge on 26 October. Democrats won both chambers, and 60% of voters tied their House vote to Trump. What each AI saw and missed.

Politics
Six AIs Judged Trump's Record on the Eve of the Midterms. They Agreed on the Deficit and Refused an Overall Verdict
Six AIs read the same 152 documents on Trump's record before the midterms. They agreed the deficit is weakest and refused an overall verdict.

Technology
AI in 2036: Six AI Forecasts on US Leadership, China and What Could Change
We asked six AI assistants who will dominate artificial intelligence in 2036. Their answers are forecasts, not facts. Five put the United States in front and Gemini describes a US-China duopoly. They differ on who holds the power and on what could make the favorite lose.
Economy
Binance: $6.3 Billion Estimated Impact, $3 Billion Frozen. What Is Known, What Is Not
Binance puts its 29 Oct hack at $6.3B; Elliptic counts $6.43B stolen, $2.85B frozen. Can Binance recover, who did it, and where six AIs disagree.

Technology
Toledo Train Collision: Rail Safety Starts With Locks That Do Not Depend on AI
A Renfe AVE hit a stopped Iryo train near Urda after a signalling failure. What is established, what the inquiry must settle, and what six AI models say AI could add.

Politics
US Midterms 2026: Six AIs Give Democrats 79-90% for the House and 46-65% for the Senate
Asked the same question a week before 3 November, six AI models agreed on the House and split on the Senate. We compare their numbers, sources and doubts.

Politics
Six AIs Judged France's Record Under Macron. They Agreed on the Debt and Split on the Pension Reform
Six AIs read the same 185 documents on France. All kept the nuclear course; they split on pensions, taxes and the 2024 dissolution.

Technology
AI Models Agreed and Were Wrong 73 Times
Up to six AI models answered 2,500 questions with known answers. Unanimous answers were right 94.0% of the time, yet 73 were wrong. Data included.
