HomeEverythingEducation
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

Web3 AI benchmark finds no top models ready for high-stakes tasks

Created at 31 May · 10:03 PM1 source↑ Market-relevant
IN SHORT

A new benchmark test of 31 AI models, including GPT-5, Claude, and Gemini, found that no system is currently prepared for the most critical tasks in Web3 AI. The study, which involved 3,543 expert questions, was presented at KDD 2026.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

31top AI models tested
3,543expert questions

Who's Involved

DMind AI
Quantified the gap in Web3 AI safety
KDD 2026
Conference where findings were made official
GPT-5, Claude, Gemini
Models included in the benchmark tests

↳ Why This Matters

The development of artificial intelligence in specialized fields like medicine (MedQA) and finance (FinBen) has progressed significantly. However, the application of AI in the Web3 space, which often involves high-stakes financial and security-related tasks, has lagged due to safety and reliability concerns. This benchmark aims to quantify the current capabilities and limitations of leading AI models in this critical domain.

Key facts

  • A peer-reviewed benchmark tested 31 top AI models, including GPT-5, Claude, and Gemini.
  • The tests covered 3,543 expert questions related to Web3 AI.
  • No AI system tested was deemed ready for the highest-stakes tasks in Web3 AI.
  • The findings were presented at KDD 2026.

The development of artificial intelligence in specialized fields like medicine (MedQA) and finance (FinBen) has progressed significantly. However, the application of AI in the Web3 space, which often involves high-stakes financial and security-related tasks, has lagged due to safety and reliability concerns. This benchmark aims to quantify the current capabilities and limitations of leading AI models in this critical domain.

Frequently asked questions

It is the first peer-reviewed benchmark designed to test the readiness of AI models for high-stakes tasks in the Web3 domain, evaluating 31 top models across 3,543 expert questions.

The benchmark included 31 top models, such as GPT-5, Claude, and Gemini.

The study concluded that no AI system tested is currently ready for the most critical tasks within Web3 AI.

The results of the benchmark were presented at KDD 2026.

What Happens Next

01Further development of AI models to meet Web3 safety requirements.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

31 May · 10:00 PM
A new benchmark tested 31 AI models, including GPT-5, Claude, and Gemini, on 3,543 expert questions, finding no system ready for high-stakes Web3 tasks.
PRN | All News Releases via PiQSuite

Sources

T1
Web3 Has No Safe AI. DMind AI Just Quantified the Gap -- and KDD 2026 Made It Official.m.piqsuite.com

Related Stories

China's AI boom spurs 'token economy' development, experts say
9 Jul · 11:05 AM
Google revamps Android AI dev benchmark, adds Fable 5 and other agents
8 Jul · 4:45 PM
Elon Musk's Grok 4.5 Launched, Priced Aggressively Against Competitors
8 Jul · 10:30 PM
Temasek to Prioritize AI Investments Over Crypto, Citing Regulatory Uncertainty
9 Jul · 10:30 AM
OpenAI, Meta, xAI Unveil New AI Models Amid National Security Review
8 Jul · 1:10 PM