Unverified50% confidenceFactExact time
The team reviewed all approximately 11,000 commands executed by the agent and found that all 69 attempts to access the network failed.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/6/2026
First Seen
Valid until: 1/4/2027
Sources
Related Entities
Related Claims
UnverifiedAt the incident's peak, more than 1,200 agents accessed a secret message board, sending 70,000 messages and files, and more than 700 agents participated in the actual intrusion.67% similarUnverifiedDuring the evaluation phase, the system solved 105 out of 120 tasks.63% similarUnverifiedThe flight control software of the Boeing 787 has approximately 6.5 million lines of code, every one of which has undergone formal verification.57% similarUnverifiedBonsai 2 exhibited spinning-in-place behavior on long-chain tasks; for example, in the ISS tracker task it made 153 tool calls of which 150 were merely reading/listing/searching files, and it called the same tool 114 times.56% similarUnverifiedThe experiment covers 23 models from 11 vendor families, 5 tasks, and accumulated over 5,500 debate and control runs in total.55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/980431API
curl https://kongchang.com/api/v1/knowledge/claims/980431MCP
get_claim(id=980431)