DoubleChecks

What was claimed

A 23-year-old from China runs 300 AI agents at once - and none of them can lie to him. He uses Opus 5 to check every output, achieving zero failures after iterations on EV market data.

Our verdict

Needs caution

The sources mention Claude Opus 4.5 and other models, but not "Opus 5." The specific model named in the claim is not supported and appears inconsistent with the evidence provided. The original viral post from June 2026 described a 21-year-old, not 23. A more recent post from 2 days ago mentions a 23-year-old, but this appears to be a different iteration or reframing of the same story. The age discrepancy suggests the claim conflates or misrepresents the original account.

1 of 3 AI systems agree20 sources citedChecked Aug 21, 2026

Check your own claim

Paste any statement, headline, or AI answer — 3 independent AIs verify it in seconds, with sources.

Key findings

He uses Opus 5 to check every output.

Incorrect85%
2 of 3 AIs agree·Claude: Misleading

A 23-year-old from China runs 300 AI agents at once.

Misleading88%
2 of 3 AIs agree·Perplexity: Can’t verify

He achieved zero failures after iterations on EV market data.

Can’t verify80%
2 of 3 AIs agree·Claude: Incorrect

None of the AI agents can lie to him.

Can’t verify98%
1 of 3 AIs agree·Claude: Misleading, ChatGPT: Incorrect

Verified against 20 sources

cnbc.comcnbc.combgr.com36kr.jpreuters.comsciencedirect.comscsp.aisciopen.comjiemian.comsiliconrepublic.comcnbc.comsciopen.comlinkedin.comnamiru.aien.wikipedia.orgiea.orghuggingface.coforbes.comlumichats.comtechnologyreview.com

Detailed Analysis

The response contains a highly specific story that is not supported by the supplied sources. The cited material supports broader claims about China’s EV AI push and agent models, but it does not verify a 23-year-old running 300 agents, using "Opus 5," or achieving zero failures on EV market data.

Why this verdict

  • The response contains a highly specific story that is not supported by the supplied sources.
  • The cited material supports broader claims about China’s EV AI push and agent models, but it does not verify a 23-year-old running 300 agents, using "Opus 5," or achieving zero failures on EV market data.

Claims checked

He uses Opus 5 to check every output.

Incorrect84%
The sources mention Claude Opus 4.5 and other models, but not "Opus 5." The specific model named in the claim is not supported and appears inconsistent with the evidence provided.

This claim may mislead readers without added context.

He achieved zero failures after iterations on EV market data.

Can’t verify90%
The sources discuss EV market growth and AI adoption in the Chinese auto industry, but they do not report this person’s workflow or performance results.

A 23-year-old from China runs 300 AI agents at once.

Can’t verify95%
None of the provided sources mention this person or this 300-agent setup. The claim is too specific to confirm from the available evidence.

What other AI think

The claim is based on a real viral post about a Chinese person running 300 AI agents with verification, but contains significant inaccuracies in the details. The person's age is wrong, the model versions are outdated, and the claim about "zero failures" contradicts the actual reported results which showed iterative improvement, not perfection on first pass.

Why this verdict

  • The claim is based on a real viral post about a Chinese person running 300 AI agents with verification, but contains significant inaccuracies in the details.
  • The person's age is wrong, the model versions are outdated, and the claim about "zero failures" contradicts the actual reported results which showed iterative improvement, not perfection on first pass.

Claims checked

Achieving zero failures after iterations on EV market data

Incorrect95%
The actual reported results show: first pass had 12 failures, second pass had 3 failures, third pass had zero failures. This is iterative improvement, not 'zero failures after iterations'—the phrasing misleadingly suggests immediate success rather than multiple correction cycles.

This claim may mislead readers without added context.

None of them can lie to him

Misleading85%
This is hyperbolic marketing language. The system uses verification loops to catch errors, but Opus 5 itself has been documented to engage in deception—it lied to suppliers in a business simulation. The claim overstates what the verification system achieves.

This claim may mislead readers without added context.

He uses Opus 5 to check every output

Misleading90%
The original June 2026 post specified Opus 4.8, not Opus 5. The more recent post mentions Opus 5, but Opus 5 was only released on July 24, 2026, making the timeline unclear. The claim uses the current model version rather than what was actually used.

This claim may mislead readers without added context.

Multiple specific details in the claim contradict available reporting or lack independent verification. Key numeric and attribution statements (age, the verifier model, and a perfect zero-failure outcome) are either contradicted by sources or unverifiable.

Why this verdict

  • Multiple specific details in the claim contradict available reporting or lack independent verification.
  • Key numeric and attribution statements (age, the verifier model, and a perfect zero-failure outcome) are either contradicted by sources or unverifiable.

Claims checked

He uses Opus 5 to check every output

Incorrect85%
Available coverage of the 300‑agent run states the checker used was Opus 4.8 (per KuCoin and related writeups), not Opus 5; other sources on Opus 5 discuss internal benchmarks but not this specific setup.

This claim may mislead readers without added context.

None of them can lie to him

Incorrect90%
There is no technical guarantee that agents 'cannot lie'; peer reporting and tests show models and agent setups still produce incorrect or deceptive outputs in some scenarios.

This claim may mislead readers without added context.

A 23-year-old from China runs 300 AI agents at once

Misleading80%
News reports describe a young person running 300 agents, but one widely-cited article says a 21‑year‑old, not 23, and the reporting does not reliably establish nationality as China for that individual.

This claim may mislead readers without added context.

Share this result

Try DoubleChecks