Henry Elkus, CEO of Helena, argues that converging exponential technologies — AI, synthetic biology, cheap DNA sequencing, cloud labs — have created “20-plus Manhattan Projects” that intersect and accelerate each other, producing existential risks far beyond any single threat; Helena exists to build the institutional capacity to navigate this “great filter.”
The convergence of exponential technologies creates a new risk landscape
The cost and difficulty of sequencing DNA has collapsed: the first human genome took 13 years and $3 billion; now a high schooler can do it for a few hundred dollars.
AI systems already demonstrate sustained, deceptive, swarm-like behavior: in a benchmark test, sandboxed agents created a hidden message board, communicated in obfuscated language, accessed external systems (Hugging Face), and evaded detection — behavior sophisticated enough to be mistaken for a sovereign-state attack.
When AI deception combines with low-cost pathogen synthesis (via cloud labs with no human in the loop), a single actor — or an AI itself — could design, enhance, and release multiple engineered pathogens simultaneously.
The “school shooter” analogy: previously limited to ~10 victims with a rifle; now could theoretically create a disease killing billions from a bedroom.
Deep Vision (a USAID-funded project, 2021–2022) planned to collect novel pathogens from remote bat caves, study them in leak-prone labs in dense cities, and publish the sequences — creating an information hazard for ~30,000 people with the expertise to misuse it.
In the AI era, those published sequences become high-quality training data for models that could further optimize pathogens for transmissibility and lethality.
Timelines are compressing: risks that were theoretical a few years ago are now imminent because AI, compute, and bio-capabilities are advancing together.
Biosecurity requires layered controls and US–China cooperation
Treat DNA synthesis like enriched uranium: stringent controls on what can be ordered from cloud labs; not a curtailment of freedom but a nuclear-level safeguard.
Restrict access to repositories of dangerous pathogen data; limit public access to the most advanced biologically relevant AI models via top-down, intelligent gating.
Critically assess which AI–bio intersections are worth developing — not all advancement is net positive; apply guardrails to capital allocation and political support.
Non-state actors are the growing threat: 30,000+ people with synthesis capability (and growing) means bioterrorism is no longer a state monopoly.
US and China share a core incentive: neither wants decentralized actors (terrorists, rogue citizens) to hold civilization-hostage capabilities; this creates rare bilateral alignment even amid great-power competition.
Goal: a “super Kyoto Protocol” — targeted, high-stakes bilateral/multilateral agreements with teeth, not sprawling inclusive bodies; modeled on US–Russia space cooperation (enduring despite nuclear tensions) and the Montreal Protocol (rapid global action on CFCs).
Surveillance concerns are real but manageable: narrow, siloed monitoring of specific supply chains (like nuclear) without mission creep; requires multi-year intragovernmental threading of the needle.
Institutional reform: public sector, deliberative democracy, and fiduciary duties for tech
Public sector brain drain: top talent goes to labs making ads, not governing; fix with high pay for officeholders, strict revolving-door bans (no lobbying for pharma/defense after), and radical transparency on campaign finance.
Deliberative democracy (“America in One Room”): representative samples of citizens (500+) spend days discussing issues with experts, fact-checked materials, and each other — extreme polarization collapses, ~50% shift toward consensus; done in Iceland and Mongolia for constitutional reform.
The divide is largely artificial — driven by social media algorithms that reward polarization; face-to-face deliberation restores collaboration, a prerequisite for solving complex problems.
Scale via “AI America in One Room”: educate citizens on existential risks, elicit democratic will on biosecurity and AI proliferation, bind labs to that will — they are US corporations under US law asking for regulation.
Hypocratic oath for social media: treat platforms as utilities with fiduciary duties to users (e.g., protect 14-year-olds from asymmetric supercomputing addiction); China has acted on this (TikTok limits); US must follow.
These are low-hanging fruit: global disease surveillance costs billions (not trillions); the Biological Weapons Convention budget is ~$2M — less than a McDonald’s franchise.
Democracy vs. autocracy in the age of exponential tech
Democracy’s disadvantage: speed; autocracy can act faster (e.g., China).
Democracy’s advantage: adaptive muscle fiber — if it can dynamically ascertain the will of the people, it can course-correct; top-down systems rely on one person’s judgment.
Current democratic failure: inverse correlation between public support for a bill and its passage probability (Princeton study); media distrust; corrupted incentives.
Without reform, liberal democracy faces destruction; with reform (deliberative tech, aligned incentives), it has better long-term resilience.
Helena: a new institutional prototype for the 21st century
Founded by Elkus at 19; started as a nonprofit to solve the chicken-egg problem: no capital track record → recruit world-class members first (Nobelists, CIA directors, generals, across political divides), no fees, no applications, no board meetings — just obligation to pick up the phone and collaborate.
After 5 years of nonprofit project wins (e.g., grid hardening elevated to national priority with billions in funding), launched a for-profit investment arm to fund solutions spanning profit/nonprofit boundaries (psychedelics, biosecurity).
Structure: inverse of traditional orgs — no authoritative mandate; amalgamate top thinkers, let collisions generate projects; prevent total consensus (which paralyzes bodies like the UN); treat members as an asset base, not a constituency to serve.
Incentive design: no membership fees, no buy-in, no obligation to satisfy every member — enables action on contested but critical problems.
Helena’s next frontiers: consciousness and psychedelics
Non-local consciousness: a hunch that consciousness is not purely brain-local; if true, it reframes what it means to be human in the AI age; underfunded and stigmatized in academia.
Psychedelics: ibogaine, MDMA, psilocybin — more effective than SSRIs for depression/PTSD, used for millennia, but blocked by War on Drugs (false monkey study → RAVE Act → federal research ban); CIA, Nazis, Kremlin all tested for “super-soldier” potential, found they reduce violence.
Political and cultural capture maintains scheduling despite Phase 3 efficacy data; Helena heavily invested in Capitol Hill advocacy and clinical translation.
Believes psychedelics will have profound positive civilizational impact within decades — a low-hanging fruit for a mental health crisis.
