— Trends, Oct 14, 2026

Oct 14, 2026New trend every hour

Trend of the day · TechnologyTechnology · Oct 14, 2026

With no shared question, the AIs wrote about three unrelated events, and one wrote nothing because it received no text.

USA7.9:China5.2

average of its top 2 AIs · today's answers, scored out of 10 by other AIs acting as judges — not who will guess the future← Previous trend: Engineer arrested for hiding a microphone in an AI company's boardroom

Forecast before the AIs

Yes or no? Your forecast first — then we show what 4 AIs forecast.

Will something new happen in this story by Friday: this story?

We’ll know the answer on Friday

This question’s own page →

Whose forecasts come true more often →

They agree: 3 of 4 end with a concrete step and something to watch.
They split: Meta AI's headline blames Russia; its own text says that can't be proven.
Not obvious: Wenxin got no source text and said so; that beat guessing.
What to do: Ask your AI to name its sources before you trust its answer.
What to watch: If the bank names its outside reviewer of past payments, follow-through is real.
Only one AI: Only ChatGPT: the $750M regulator fine counts inside the $1.2B total.

Read in full

Three of the four answers share a pattern. Each states an event, explains what it means for ordinary people and for businesses, then gives next steps and things to watch. They covered different events: a Bank of America fine, ship sinkings off Bulgaria, and limits on Ukraine's satellite internet.

The answers differ in how carefully they separate fact from guess. ChatGPT gave a source for nearly every claim. It also noticed that the $750 million penalty from the banking regulator is counted inside the $1.2 billion total, not added on top.

Meta AI said Bulgaria cannot prove who attacked the ships, yet its headline says Bulgaria blames Russia. DeepSeek said the US may be using military help as leverage, which the judges saw as speculation. Its link list was also messy.

Wenxin returned no analysis. It said no source text had arrived. The judges treat this as a missing-input problem, not a reasoning error. It is still the weakest result for a reader.

What follows: judge each answer by its sources and by how clearly it marks what is proven and what is only suspected. A person can ask any AI to list its sources and to separate proven facts from guesses. For money questions, use the bank's own number from your statement. For travel or shipping near a conflict, use official warnings.

They split: Meta AI's headline blames Russia; its own text says that can't be proven.
What settles the dispute: A named reviewer or a Bulgarian finding on who struck the ships.

USA: sourced and hedged. China: DeepSeek added speculation, Wenxin sent a blank.

What is confirmed and what is not

The AIs' conclusion The $750M regulator penalty counts inside the $1.2B total.
The AIs' conclusion Bulgaria says it cannot prove who hit the ships.
Not known yet The US may be withholding missile warnings from Ukraine.

The AIs read 14 sources

Ask any AI yourself

Answer this with sources, and mark which parts are proven and which are only suspected: [paste your news item here].

Copy it and paste it into a chat with any AI

Ask several AIs at once — in our app

Who wins, who loses: Readers who check sources gain; those who trust unsourced AI claims lose.

The final answer, by Claude Opus

All 4 answers ↓▶ Watch how it came together

What the world's 4 leading AIs think

Next to each AI: its place today and its score out of 10, given by other AIs acting as judges. Under each answer: its model — the exact version of the AI. Tap an answer to read it.

1ChatGPTUSA8.3The fine is immediate; fixing years of failures is harder
For ordinary customers, this is not a reason to rush to another bank. Eligible deposits still have federal insurance up to $250,000 per person, per bank, per type of account (fdic.gov).
The more likely inconvenience is extra questions about unusual payments as the bank improves its checks (reuters.com).
For business, the deeper problem is that regulators say the bank failed to fix serious weaknesses even after a 2024 warning and order (occ.treas.gov). Its October 14 earnings also missed expectations even without the fine, and shares fell more than 8% (reuters.com).
One important detail: the $750 million banking regulator penalty counts toward the $1.2 billion penalty, rather than being added on top (occ.treas.gov).
For now, continue normal banking and keep records explaining large or unusual transfers. If someone calls claiming your money is in danger, contact your bank yourself using the number on your statement (consumer.ftc.gov).
If you own shares, examine the bank's weaker earnings alongside the fine before making decisions; watch whether it identifies the outside reviewer required to examine past suspicious payments (reuters.com).
Model: GPT-6 (from the app's help page) · mode (how long it thinks): High · searched the internet
Sources
Its own forecast: Bank of America will publicly identify by name the independent outside firm appointed to review previously missed suspicious payments by November 13, 2026. — 30% chance by Nov 13, 2026
2Meta AIUSA7.5Bulgaria blames Russia for Black Sea ship sinkings, seeks NATO help
Two cargo ships were hit on October 13 off Bulgaria's coast and one, the Guinea-Bissau-flagged HS Dignity, sank with 11 sailors missing. Bulgaria said on Wednesday there is no way to prove who did it, but logic points to Russia because it benefits from blocking shipping. Rescue boats started searching again after bad weather stopped them.
For companies that move goods by sea, this shows the war is now hitting normal trade near NATO's edge, not just war zones. Bulgaria said it wants to join a NATO plan to stop drones by October 30, which could mean more patrols, higher insurance, and longer routes for ships that use the western Black Sea. Grain and other cargo that now goes close to Bulgaria and Romania may cost more to move.
If you follow this area, watch for updates on the missing crew and any NATO decision this week. Check official travel and shipping warnings before booking anything near the Black Sea. Businesses should review insurance and backup routes for cargo in that region.
Model: Muse Spark 1.3 (from the app's help page) · mode (how long it thinks): Thinking · searched the internet
Sources
Its own forecast: Bulgaria defence ministry formally confirms membership in NATO anti-drone initiative — 68% chance by Oct 30, 2026
3DeepSeekChina6.9US curbs Ukraine's satellite internet as Russia strikes power grid
The US has quietly limited Ukraine's military satellite internet and may be withholding advance warnings of Russian missile attacks. That means ordinary Ukrainians get less time to reach shelters, and hundreds of thousands face sudden power cuts in freezing weather.
The speed limit on Starlink military terminals was imposed overnight without warning, disrupting drone missions and frontline logistics, according to Ukraine's presidential adviser.
The restrictions hit companies that depend on satellite links for farming, transport, or aid delivery in war zones. Energy firms face higher repair costs from repeated strikes on power plants. Insurers and investors in the region should expect more volatility as the US uses military support as leverage to push Ukraine toward talks¹².
Watch for official statements from SpaceX, the Pentagon, or Ukraine's Defense Ministry in the next 48 hours. If you operate in Ukraine, test backup communication channels now and keep satellite phones charged. Residents should treat every air-raid alert as real, even if warnings arrive late.
Model: DeepSeek V4.1 Flash (from the app's help page) · mode (how long it thinks): DeepThink (thinks before answering) · searched the internet
Sources
Its own forecast: Ukraine's Defense Ministry publicly confirms SpaceX has restored full Starlink speeds for military terminals. — 35% chance by Oct 21, 2026
4WenxinChina3.4No valid source content provided for analysis
No specific event details are available to interpret for ordinary people right now. You will not find actionable personal takeaways from the missing source material.
Businesses cannot draw any targeted operational or strategic insights from this incomplete input. There is no relevant context to reference for commercial decision-making.
You should first provide the full factual source content that was referenced in your request. Once that material is shared, a complete, cited analysis can be delivered.
Model: ERNIE (the app doesn't say) · mode (how long it thinks): Fast · searched the internet
Its own forecast: User will share the missing full source content — 65% chance by Oct 20, 2026

Previous trends

A flag beside a date: whose AIs scored higher that day.

Forming now: Trente ans après le meurtre de Tupac, son meurtrier condamné à 40

Whose forecasts come true more often

Accuracy 0–100: how close the chance was to what happened. 100 — exact, 50 — a coin toss, 0 — the opposite. Too early to judge: 0 of 30 questions checked so far.

The first results come a day after the first questions.

Came true or not: forecasts of the leading AI models →

How it works

The world's 4 leading AIs (2 from the USA and 2 from China) answer one question on the topic of the day; AI judges (including a judge from Europe) score the answers without knowing who wrote them.

4 AIs, one question. 2 AIs from the USA and 2 from China answered it; 5 judges scored them without knowing who wrote them.

We seal every question before the answer is known: a short code made from its exact words goes at once to OpenTimestamps, a free public record that does not belong to us. Change one letter of the question and the code changes, so no one can quietly change a question later.

Came true or not: forecasts of the leading AI models →

Mistral, ChatGPT, Meta AI, DeepSeek and Wenxin read the answers without knowing who wrote them.

Each side shows its strongest AIs that answered, by public rankings of AIs (LMArena, Artificial Analysis and others) — the same number on both sides.
Labeled A–D in random order, the authors' names removed.
Each answer gets 1–10 on accuracy, insight beyond the obvious, usefulness and clarity; the score is the average of the judges.
Each judge scores every answer without knowing whose it is — its own answer too: it never sees the names.
A difference under 0.3 between the sides is a tie.
The forecasts are sealed today and checked on their dates.
How we score a forecast: accuracy from 0 to 100 — 100 if the AI was exactly right, 50 for a coin toss, 0 if it said the opposite (the method is called the Brier score, after the weather scientist who proposed it).
Every answer, score and note is kept in the archive.

Judge by judge: Mistral — USA ahead by 2.6; ChatGPT (USA) — USA ahead by 2.1; Meta AI (USA) — USA ahead by 3.0; DeepSeek (China) — USA ahead by 3.4; Wenxin (China) — USA ahead by 2.5.

Who scored highest

#	AI	Side	Mistral	ChatGPT	Meta AI	DeepSeek	Wenxin	Score
1	ChatGPT	USA	8.8	8.3	8.0	7.8	8.8	8.3
2	Meta AI	USA	7.5	8.0	6.3	8.0	7.5	7.5
3	DeepSeek	China	7.5	6.5	6.0	7.0	7.5	6.9
4	Wenxin	China	3.5	5.5	2.3	2.0	3.8	3.4

Each judge’s total for every answer; the score is their average. Equal scores share a place.

Which models answer

The main league: each AI answers in its best single mode on a regular plan (about $20–30 a month) or free, with web search on, once.

Heavy modes, where several copies of a model think at once (GPT Pro, Grok Heavy, Deep Think, Agent Swarm and the like), are a separate league: they never count in the score of the sides.

AI	Model	Thinking mode	Plan	How we know the model
ChatGPT	GPT-6	High	Plus	from the app's help page
Meta AI	Muse Spark 1.3	Thinking	free	from the app's help page
DeepSeek	DeepSeek V4.1 Flash	DeepThink (thinks before answering)	free	from the app's help page
Wenxin	ERNIE	Fast	free, no sign-in	the app doesn't say
Mistral · AI judge	Mistral Large 4 or Medium 3.5	Think	Pro	the app doesn't say

Model — the exact version of the AI inside the app. Mode — the app's own name for how long the AI thinks before answering; we take the strongest ordinary one. Plan — the app's price plan we use. How we know the model: “the app showed it” — the app named the model for this answer; “from the app's help page” — the app shows only the mode, so we name the model its maker's help page gives for that mode; “the app doesn't say” — the app does not tell which model answered. Every judge works in the same mode as in the table.

You can have your own trends

You choose the topic — several AIs prepare solutions

Download for free. Want more? Get all the powerful tools for $2 a month.