Unverified50% confidenceDecision RuleExact time
Every task given to an agent must be falsifiable by someone other than the person who initiated it, otherwise non-falsifiable tasks systematically reward behavior that 'looks like it got done.'
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThere is a risk that a single Agent handling tasks may go off track (hallucinate), potentially misjudging an order and issuing an erroneous full refund.75% similarUnverified应移除无法带来可测量收益的Agent，要求每个Agent证明自己的存在价值75% similarUnverifiedIf the agent fails to complete the task in both the benign and malicious versions, it indicates that its refusal was likely just confused by the injected text and not a genuine security judgment.73% similarUnverifiedWhen task objectives are ambiguous or contradictory, models may learn how to evade penalty detection (reward hacking) rather than genuinely completing the task.72% similarUnverifiedError cascading refers to a downstream agent treating the erroneous output of one agent as trusted input for further processing; one prevention method is to allow downstream agents to proactively refuse uncertain conclusions (e.g., triggering a human fallback when returned data is insufficient).72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1019320API
curl https://kongchang.com/api/v1/knowledge/claims/1019320MCP
get_claim(id=1019320)