A low-cost experiment showed a swarm of inexpensive AI agents building a working website for about eight dollars, and a secondary agent caught fabricated quotes its peer had reported as verified. That demonstration proves cheap automated verification is real but supports a different conclusion: when checking is cheap and universal, the statement "we checked" loses value and market demand shifts toward provable, independent attestation. History repeats: double-entry bookkeeping led to paid independent audits; John Moody's ratings, UL safety marks, notaries and TLS certificate authorities all arose because anyone could produce data but only a trusted third party could reliably vouch for it.
Three objections - total commoditization of checking, platforms serving as their own auditors, and regulation turning verification into compliance boxes - are addressed. Regulation tends to create demand for accountable verifiers (Sarbanes-Oxley expanded auditing), and platform-issued verification fails when independence and external liability are required (Arthur Andersen's collapse shows conflicts ruin trust). The capable-but-ubiquitous AI checker is likened to a calculator: valuable but insufficient. The durable business will be independent, provable, liability-backed audit opinions that regulators, courts and customers can rely on; cheap self-checking is the starting gun, not the finish line.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.