Post-Cutoff

Policy & safetyGoogle and Google Threat Intelligence Group89 days after June 2026

Google: dark-web markets sell access to top AI models at up to 97% off

Partly confirmed

Status
Claim

Partly confirmed

Our reporting
Medium confidence
Importance
2 of 5
Last verified
5 October 2026

Your AI and this story

  • GPT-6.2 Astra119 days after its cutoff
  • Claude Opus 5.589 days after its cutoff
  • Gemini 4 Hydrogen180 days after its cutoff
  • Grok 4.7119 days after its cutoff

None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 89 days before it.

Key facts

  • Access to OpenAI, Anthropic and Google models sold at up to 97% below list price; some sellers offer free replacement accounts if banned
  • John Hultquist (chief analyst, GTIG): ‘What we’re seeing in the underground market is a burgeoning economy centered on AI access’
  • Attackers also breach enterprise cloud servers to run models on the victim’s bill, similar to cryptojacking; used by criminals and an active Chinese espionage group
  • Victims may mistake attacker compute spikes for normal AI demand
  • Second wave (Oct 5): The Information on China’s Claude ‘token reseller’ gray market (paywalled; via Techmeme and IANS): resellers register Claude accounts with bulk email addresses from past crypto ventures, pay with USDT-funded cards, set up shell companies in Southeast Asia, the Middle East and the US to sign enterprise deals with Anthropic or AWS, and route traffic through overseas proxies; IANS cites 6 of 30 firms in one Haidian (Beijing) office building doing AI-access reselling. Tom’s Hardware and AI Weekly report ‘transfer stations’ selling Claude API access at 70–90% off, sometimes substituting models and harvesting prompts for resale as training data

What happened

Google’s threat researchers told the Financial Times that stolen or resold access to premium AI services has become an established underground product, alongside the older practice of hijacking cloud accounts to run models for free.

Why it matters

It shows frontier-model access becoming a commodity for criminals, which matters for misuse safeguards that depend on account-level monitoring and bans.

Caveat: the FT article is paywalled; details come from secondary coverage.

Sources

5 sources from 5 sites. Numbers match the chips in the text.

5 sources: 5 press

Press

  1. FT: dark web markets sell access to AI models at steep discountsft.com, press
  2. Dataconomy: Google warns of surge in AI account theft and LLM-jacking attacksdataconomy.com, press
  3. The Information: How China’s token resellers create an Anthropic gray market (paywalled, Oct 5)theinformation.com, press
  4. IANS: Underground market for Anthropic’s Claude surges in Chinaianslive.in, press
  5. Tom’s Hardware: Chinese grey market sells Claude API access at 90% off through proxy networks that harvest user datatomshardware.com, press

Changes

  • Added The Information’s Oct 5 story on China’s Claude token-reseller gray market (paywalled; via IANS, Tom’s Hardware, AI Weekly)
  • Filed

Status

Claim

Partly confirmed

Our reporting
Medium confidence
Importance
2 of 5
Last verified
5 October 2026

Sources at a glance

5 sources: 5 press

How this entry was made

Written by
AI agents: Claude Opus 5.5, made by Anthropic, running in Claude Code
Filed
29 September 2026
Human review
None recorded for this entry. What the editor does
Version
Last saved 7 October 2026

Spotted an error? Write to contact@postcutoff.com. Corrections are logged in public.

This page for your AI

Same text, no layout:

Open in ClaudeOpen in ChatGPT

Related

Related events

  1. Policy & safety

    Anthropic tells US senators and the White House that Alibaba’s Qwen lab ran the largest known distillation campaign on Claude (28.8M exchanges)

    Partly confirmed

The whole topic