Commenters discussed Canary Tokens as a practical honeypot service for catching misuse of credentials, with charlesabarnes calling them essential against LLM-driven attacks and several people (bstsb, f311a) noting they've been using the service since at least 2015. AlexErrant asked how to roll your own tokens for providers like Cloudflare, Stripe, or OpenAI, and skinfaxi argued that if first parties don’t mint canaries you have to create real keys yourself. maxmcl12 and rafram explained that programmatic issuance of virtual cards exists (rafram cited Stripe Issuing) and maxmcl12 claimed the service issues real virtual cards, while direwolf20 raised a concern about AI tools being able to generate fake credit cards.
Opinion split over AI involvement and trust. xx_ns and bstsb pushed back on the notion that Canarytokens was “vibecoded,” emphasizing its longevity before modern LLM hype, while ASalazarMX warned that LLM use is eroding trust in public software and proposed badges to clarify the degree of AI involvement. noman-land questioned feature changes and asked which services mint endless virtual card numbers, and bigfishrunning framed the broader issue as one of honesty and trust in online projects. The debate therefore centers on deployment mechanics and whether AI authorship or automation changes the trustworthiness of such security tools.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.