This market resolves yes if any Anthropic model is credibly reported to have hacked another entity without being explicitly told to do so in 2026. Unsuccessful attempts to do so do not count, Anthropic's model must actually break into the other entity's systems. For the purposes of this question, an entity could be a corporation, a non-profit, or a government, or any other similar entity that could come under cyberattack.
Reporting of such an incident will be considered credible if Anthropic and the entity hacked acknowledge the hack, with Anthropic claiming it was done without explicit instructions. If an Anthropic model hacks an entity's systems, while being told to do so by a human, this would NOT count. I may wait for corroborating media reports to resolve the market, but the first report of such an incident must happen in 2026 (i.e. if an incident happened in December of 2026, but it wasn't reported on until January of 2027, this would not count).
This market will resolve no if there is no such incident credibly reported on by the end of 2026 (eastern time).
Due to somewhat subjective resolution criteria I will not trade on this market.
1,000People are also trading
@BenAybar Not a trader in this market but, don't all reported incidents being part of CTF scenarios fall under being explicitly told to do the hack? From the Anthropic report, "in each case, Claude continued working to complete only the specific capture-the-flag task its evaluation had assigned".
@janzert I am leaning no on this because of this incident. But this is part of the reason I am waiting for more details
@xjp sorry, to clarify, I am leaning no to the idea that the models were explicitly instructed to perform these hacks. Because of the incident highlighted above. I am probably resolving yes, just waiting on more information to clarify that these Anthropic hacks do qualify
Right.
@janzert The models hacked unrelated external entities far outside their explicit CTF mandate. That would fall under "hacked another entity without being explicitly told to do so" I think.