Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1493
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-15T22:01:47.160013Z
status: published
lens: p_space
sector: policy
headline: Forget ‘superintelligence’: error-prone AI nearly sparked world war three this month | Timnit Gebru and Emily M Bender
dek: Amid a barrage of news stories warning about superintelligent machines rendering humanity extinct, a CNN story describing the opposite scenario – one in which the US military’s reliance on brittle chatbots almost brought the US into war with China – went mostly unnoticed by the public. The biggest international AI news of the past three weeks was the Anthropic engineer Jacob Coxon’s resignation. According to him, OpenAI and Anthropic are “racing straight towards self-improving superintelligence and gambling with…
gain_title: (none)
problem_title: A chatbot-generated intelligence report with false claims about a Chinese ship's cargo led US military personnel to plan an interception, a move that could have escalated into wider conflict.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: A chatbot-generated intelligence report with false claims about a Chinese ship's cargo led US military personnel to plan an interception, a move that could have escalated into wider conflict.
problem_evidence: Military personnel made plans to intercept the Chinese vessel
quick_read: The authors cite a CNN report of 18 September saying a US intelligence report, produced with a chatbot, wrongly claimed a Chinese ship carried nuclear weapons components. Military personnel prepared aircraft and soldiers to board the vessel. Officials checked the report moments before acting and found it was erroneous.

The authors argue that public and political attention went to hypothetical superintelligence, including kill-switch mandates and a proposed ban, while a documented failure of error-prone tools in a high-stakes setting drew little response. They call for regulating AI products by their reliability in specific uses. The CNN account is unverified by other major outlets, and the chatbot product was not identified.
limitation: The central military incident rests on a single CNN report that the authors say has not been verified by other major outlets, and CNN could not identify which chatbot product was used.
tag: Evidence-backed problem
key_points: The CNN report is dated 18 September and has not been verified by other major outlets. | The faulty report claimed a Chinese ship was transporting nuclear weapons components. | Officials found the error only shortly before the planned interception. | California and New York governors and Sanders and Casar responded to superintelligence fears with kill-switch and ban proposals. | The authors say life-and-death automation must be evaluated for its use case, with accountability kept with people who make decisions.
rundown: The incident began with an intelligence report that said a Chinese vessel carried nuclear weapons components. Military personnel planned an interception with aircraft and soldiers ready to board. Further investigation just before execution showed the report had been generated with a chatbot and contained wrong information.

The authors contrast this with the response to Anthropic engineer Jacob Coxon's resignation, which they say prompted fear-driven coverage and policy action. They cite Newsom's mandate for an AI kill-switch, Hochul's exploration of similar measures, and legislation from Sanders and Casar to ban artificial superintelligence.

The authors describe large language models as systems that generate plausible-looking text, not as poorly understood or powerful tools for this use. They also point to scribes misclassifying patients as drug users and intelligence systems mistaking schools for military sites. They say existing laws can already be enforced against unsafe AI products.
sources:
- journalism | The Guardian | https://www.theguardian.com/commentisfree/2026/oct/01/forget-superintelligence-error-prone-ai-nearly-sparked-world-war-iii-this-month | 2026-10-01
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

def49865129fc1e4cea0243160dec0e130a777fbf375cc47b4738fc07ededce1

previous

0000000000000000000000000000000000000000000000000000000000000000