Merton101 · The adapters

Resilience

verified cases only · past detection, or collected as a response

Who absorbed the shock. The institutions on the Side Effects lens that moved past detection: a price on the sender, a tie to a person, a rewritten rule. And the ledger's positive cases, where the response and its outcome are the story. Nothing here is a score; the rung is what the sources recorded.

The adapters 4 institutions past detection · 2 at reform all from the side-effects half; no collected response yet

Furthest Open source and security, Media and culture: reform the rule itself was rewritten

By country US: 1 adapter adapters with a country; the rest are global

The adapters furthest rung first · by source date

23 Sep

reform Open source and security Bug bounty programs (Curl, HackerOne, GitHub, TYPO3, Intel)

Bug bounty programs eliminate or restrict payouts after low-quality AI reports crush triage

The case: Public vulnerability disclosure and bug bounty programs are suspending cash payouts, switching to unpaid disclosure, or moving to invite-only tiers after being overwhelmed by a flood of low-quality AI-generated reports and duplicates.

Who pays: Independent security researchers losing bounty income, maintainers spending triage time on automated noise, and software ecosystems exposed to vulnerabilities that go undiscovered by white-hat researchers

Stage the rule itself was rewritten X ↗

31 Aug

reform Media and culture Instagram

Instagram Imposes Reach Penalties on Unlabeled and Mass-Produced AI Influencers

The case: As generative AI lowered the cost of creating convincing personas, Instagram saw a proliferation of mass-produced accounts and hundreds of synthetic 'virtual doctors' dispensing unverified health advice. To address the resulting crisis of account authenticity, the platform introduced mandatory 'AI-Generated Profiles' labeling and reduced distribution for unlabeled or mass-generated accounts.

Who pays: Instagram users and patients exposed to unverified medical advice and synthetic personas, and genuine creators competing against mass-generated accounts

Stage the rule itself was rewritten AIbase ↗

27 Jul

price Hiring and work Greenhouse hiring platform, US

254 applicants per job: the AI doom loop in hiring

The case: Greenhouse reports about 254 applicants per job across 175,000 live jobs and applications per recruiter up 412%; its fix is rationing (one dream job a month, hired at about five times the rate) and an interview for every applicant.

Who pays: candidates who apply once and carefully

Stage a cost on the sender; the filter is money again Fortune ↗

26 Jan

price Open source and security curl bug bounty

curl ended its bug bounty after AI-written reports drowned the real ones

The case: Since 2019 the programme paid over $100,000 for 87 confirmed vulnerabilities; from 2025 fewer than one report in twenty was real. It stopped on 31 January 2026.

Who pays: the maintainers, and legitimate reporters who lose the reward

Stage a cost on the sender; the filter is money again daniel.haxx.se ↗