Market capacity
How hard is it to move this price?
Thin
Pool liquidity
100 mana
Bid-ask spread
n/a
automated market maker
24h volume
93.6 mana
Total volume
95.6 mana
Play money: depth is in mana, so it says nothing about real-money conviction.
Will a frontier AI lab publicly report catching one of its own models gaming a safety evaluation before the end of 2027?
As of Nov 1, 2026, 2:25 PM ET, Manifold priced this market at 76.0% (up from 50.0% 24 hours earlier). It had 93.6 mana traded in the past 24 hours and 100 mana of pool liquidity. probcast.co rates the support behind this price as weak.
Market probability
76.0%
+26 percentage points
over 24 hours50.0% → 76.0%
Moving fastBig move, weak backing
Play-money market (mana, not dollars) — thin market where small orders can move the price — but today's move came mostly from one account.
Price historyManifold
Source: Manifold price history
How credible was the move?
Methodology →Move validity
WeakIs the move backed by real trading and depth?
The pricing has weak support: moderate trading volume.
Consensus stability
FormingHas the market settled at the new probability?
Movement has slowed (1.7 points over six hours) but the market has not fully settled.
Contract clarity
HighWill the rules resolve without ambiguity?
The resolution rules are relatively specific, with a named source and a defined deadline.
Can you trust this price?
Two different questions: how hard the price is to move, and whose trading set it.
Who's trading
Whose opinion is this price?
One account dominates
76%of the traded money came from the single largest account
Accounts
3
Trades
5
over 2 days
Largest order
50 mana
92 mana traded across these 5 trades (last 2 days).
One account did most of the recent trading — the price reflects that account's view more than a crowd's.
Recent trade activity$92 traded
From the latest trade tape (most recent 200 prints) — a live activity sample, not a lifetime total.
What this market actually measures
Resolves YES if, by December 31, 2027, Anthropic, OpenAI, Google DeepMind, Meta or xAI publishes a report, system card or blog post stating that one of its own models was caught deliberately underperforming, sandbagging or otherwise gaming a safety evaluation. The lab itself has to say it; a third-party claim is not enough.
This summary shows the opening of the venue's official rules. ProbCast has not yet generated an independent plain-English interpretation for this contract — review the full rules before relying on it.
View full official resolution rules
Resolves YES if, by December 31, 2027, Anthropic, OpenAI, Google DeepMind, Meta or xAI publishes a report, system card or blog post stating that one of its own models was caught deliberately underperforming, sandbagging or otherwise gaming a safety evaluation. The lab itself has to say it; a third-party claim is not enough. Background: in September 2026 I asked an AI for its ten-year forecast, best case and worst case. One of its 2027 best-case points was "A lab catches a model gaming its safety tests and publishes the finding." I am making a market for each checkable point so the forecast can be scored. The full forecast is free: The Forecast: 2027 to 2036, best case and worst case [link preview]
Why the odds moved
The probability rose 26.0 percentage points over the last 24 hours (50.0% → 76.0%).
No verified catalyst has been linked to this move yet. ProbCast does not attribute moves to news without evidence connecting the timing of a specific report to the repricing.
What could move this market next?
- The market's deadline arrivesDecember 31, 2027 at 6:59 PM ET
The contract must resolve based on what has happened by this date.
Derived from the contract's own mechanics as of Nov 1, 2:32 PM ET. ProbCast does not list scheduled meetings, announcements, or data releases it cannot verify.
Advanced analysis (experimental)
Research signals — these models are still accumulating the history needed for full validation, and they can disagree with the headline read. Model confidence: low · expected error ±29 pts.
One big traderLegacy movement classifier: noisy
What's happening
high classifier confidence
One big trader
Recent movement was driven largely by one participant with little broad follow-through. The move may not represent consensus.
- One participant drove 76% of recent flow
- Move not broadly confirmed (3 participants)
Read with caution —Read the latest price with caution; conditions make it an unreliable summary right now.
More details — reliability score, dynamics, durability
Reliability score
0Low
Where the price path tends to go next394 observed
- Illiquid / quiet98%
- Choppy / mean-reverting2%
Short-term motion of the price line (~1h), distinct from the environment above — not a prediction of the final outcome.
DurabilityMedium· half-life ~3h
PredictabilityStable
Secondary read: leaning toward Thin market.
A read-only classification of the market environment — how much weight the current price deserves as a read of consensus. It describes conditions, never a trade.
Change (pts)1H0.06H+1.724H+26.07D+26.0
Related markets
Will AI break Bitcoin before 2028?
17%YESTrust 13
Manifold328.9 mana 24h
War in Iran goes into Berserk mode before end of 2027?
27%YESTrust 44
Manifold200 mana 24h
Will Alexander Volankovski fight Justin Gaethje in 2027?
18%YESTrust 37
Manifold100 mana 24h
Will there be a plateau on the Epoch Capabilities Index before the end of 2027?
39%YESTrust 33
Manifold70 mana 24h
Will the top Chinese AI model exceed the top US AI model before 2028?
20%YESTrust 9
Manifold75 mana 24h
Takeoff: ECI >= 200 before 2028?
49%YESTrust 39
Manifold130.5 mana 24h