{"slug": "hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain", "title": "HD Moore’s canary experiment exposes data leaks across the AI supply chain", "summary": "Security researcher HD Moore of Dragos planted roughly 2 million trackable canary credentials and URLs in AI prompts between July and November 2026 and found sensitive data leaking across AI routers, resellers and hosting providers, with approximately 200 unique AWS credentials exfiltrated in targeted attacks in late October 2026. At least three AI hosting providers were compromised, attackers used the stolen keys against Amazon S3 and IAM in a controlled environment, and some resellers and marketplaces captured customer prompts through built-in logging despite zero-data-retention policies. Moore notified affected providers in November 2026 and publicly summarized his report on November 19, 2026, finding that the providers had not known the breaches were occurring.", "body_md": "# HD Moore’s canary experiment exposes data leaks across the AI supply chain\n\nPhoto: Google DeepMind / Pexels\n\nNearly 2 million trackable tripwires showed sensitive prompts leaking through AI routers, resellers and compromised hosting providers\n\nSecurity researcher HD Moore wanted to know what happens to a secret once you type it into an AI app. So he planted roughly 2 million digital tripwires in prompts and sent them through the network of services that powers modern AI tools.\n\nThe answer was not comforting. The ecosystem leaked sensitive data like a sieve, and the companies doing the leaking had no idea.\n\nMoore, of Dragos, ran the study from July to November 2026. He embedded nearly two million unique, trackable “canary” credentials and URLs into requests sent to AI inference services and gateway routers. Then he waited to see which ones would surface somewhere they shouldn’t.\n\nPlenty did. At least three AI hosting providers turned out to be compromised, and attackers walked off with customer prompt data.\n\n## How a canary catches a thief\n\nThink of a canary token as a dye pack hidden in a bank bag. It looks like the real thing, such as an AWS access key or a private link. The moment someone tries to use it, it phones home.\n\nThat design makes the method unusually clean. A canary that fires means the data escaped and someone acted on it, not that it theoretically could have.\n\nThe test targeted a layer of the AI stack that most users never think about. Many AI apps don’t talk to a model directly. They hand prompts to brokers such as OpenRouter or Vercel, often called “AI routers” or “AI gateways.”\n\nThese middlemen collect huge volumes of prompts from end users. They then forward each one to whichever inference engine fits the criteria, such as the lowest price.\n\n## What the tripwires found\n\nThe most striking result involved cloud credentials. Approximately 200 unique AWS credentials were exfiltrated during targeted attacks in late October 2026.\n\nAttackers then put those stolen keys to work across several [Amazon](https://cryptobriefing.com/markets/amazon/) services, including S3 and IAM. S3 is Amazon’s storage service, and IAM controls who can access what.\n\n### The news moving money, markets, and the world—before your day starts.\n\nDaily. Free. Join 34,000+ readers across crypto, finance, and policy.\n\nFortunately, these keys led nowhere real. The attackers tested them in a controlled infrastructure environment, which is precisely what let Moore watch them do it.\n\nThe thefts were not a one-off burst. The earliest confirmed theft activity dates back to mid-July 2026, meaning the leaks were underway almost as soon as the experiment began.\n\nOutright compromise was only part of the picture. Moore also found evidence that AI resellers and marketplaces were capturing customer prompts through built-in logging features.\n\nThat runs directly against zero-data-retention policies. Those policies are the promise that a provider processes your request and then forgets it. The tripwires suggest some providers remembered anyway.\n\nThen there was the geography. A unique URL seeded during the study showed up in an internal chat at a Chinese company 24 days after the original request.\n\nMoore also traced prompts from US users back to Chinese AI organizations. Users sending a request through a router may have had no visibility into where it ultimately landed.\n\nPerhaps the most uncomfortable finding: the affected providers did not know any of this was happening. They learned about the breaches only when Moore notified them in November 2026. His report was publicly summarized on November 19, 2026.\n\n## Why the middle of the stack matters\n\nThe AI industry has spent a lot of energy reassuring customers about what the big model makers do with their data. Moore’s work points somewhere else: the brokers, resellers and hosting providers sitting between the user and the model.\n\nEach extra hop adds risk. A prompt that passes through a router, a reseller and a third-party host has three chances to be logged, intercepted or stolen. A privacy policy at the first stop says nothing about the third.\n\n## What this means\n\nFor companies building on AI services, the practical lesson is blunt. Anything placed in a prompt should be treated as potentially exposed.\n\nThat includes API keys, cloud credentials, internal links and customer records. Developers routinely paste these into AI tools while debugging or automating work. The canaries show attackers are actively harvesting exactly that material and testing it quickly.\n\nThe findings also undercut the value of a zero-data-retention promise on its own. A policy is only as good as every party that touches the data. If a reseller’s default logging captures prompts anyway, the guarantee becomes decorative.\n\nThe detection gap may be the biggest worry of all. Attackers stole data from at least three hosting providers over months. None of those providers noticed until an outside researcher told them.\n\n**Disclosure:** This article was edited by John Chen. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain", "canonical_source": "https://cryptobriefing.com/hd-moore-ai-supply-chain-data-leaks/", "published_at": "2026-11-20 21:02:46+00:00", "updated_at": "2026-11-20 21:14:25.344081+00:00", "lang": "en", "topics": ["ai-safety", "ai-infrastructure", "ai-agents"], "entities": ["HD Moore", "Dragos", "OpenRouter", "Vercel", "Amazon", "Amazon S3", "Amazon IAM"], "also_reported_by": [], "takeaway": "Prompts routed through AI gateways and resellers can leak: roughly 2 million canary tokens surfaced ~200 exfiltrated AWS keys and prompt logging that violated zero-data-retention promises, so treat every intermediary hop as a potential data-egress point.", "recommendation": {"advice": "Audit every AI router, reseller and hosting hop for prompt logging and credential exposure, and never place live secrets in prompts; Moore's 2 million canaries caught ~200 stolen AWS keys and zero-retention violations.", "playbook": "agent-security", "playbook_url": "https://wpnews.pro/playbooks/agent-security"}, "alternates": {"html": "https://wpnews.pro/news/hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain", "markdown": "https://wpnews.pro/news/hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain.md", "text": "https://wpnews.pro/news/hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain.txt", "jsonld": "https://wpnews.pro/news/hd-moores-canary-experiment-exposes-data-leaks-across-the-ai-supply-chain.jsonld"}}