Roman Yampolskiy dropped a mystery on Peter McCormack's podcast. We went hunting.
Yampolskiy — AI safety researcher, P(doom) ≥ 99% — told McCormack about a viral 2016 story he couldn't verify: two tech billionaires hiring scientists to break us out of the simulation. The story disappeared. We found the source, traced the money, and identified the most likely names. Then we hit the wall.
▶ youtube.comWhat Bitcoin Did · April 202696 min
The Catalyst~14:30 mark
Maybe 10 years ago in all the media articles, there was a story that some billionaires hired a team of people to hack them out of a simulation. It was going viral and then it disappeared. I cannot find the source. I talked to people who should know and they kind of acknowledged knowing but there is zero info.
AI & EMPIRE · SIMULATION HYPOTHESIS · FOLLOW-THE-MONEY
DOSSIER YAMPOLSKIY-SIM-ESCAPE · 2026-05-13
🧠
🚀
💻
🎓
🤖
🏛️
Yampolskiy's Missing Source: Tracing The "Two Billionaires" Who Allegedly Tried to Hack The Simulation
A New Yorker passage from October 2016 named "two tech billionaires" funding secret simulation-escape research. The originating claim is CONFIRMED. Identities are inferred: Musk at ~90%, Tallinn at ~40-50%. The institute that most plausibly hosted the work was shut down by Oxford in 2024. No insider has spoken publicly in ten years.
In an April 2026 podcast interview, Roman Yampolskiy did something rare in AI safety: he named a story he could not verify. "Two billionaires hired a team of people to hack them out of a simulation. It was going viral and then it disappeared. I cannot find the source." We watched the video. We pulled the contemporaneous Hacker News thread, the Wayback snapshot, the institutional reports. The viral story is real — Tad Friend's October 10, 2016 New Yorker profile of Sam Altman contains the exact passage. The originating claim is verified. The identity claim is inferred: Musk's £1M to FHI Oxford is on the public record (~90% probable as one of the two); Tallinn's role is circumstantial — institutional proximity, ideological fit, and post-Friend podcast adjacency (~40-50%, not a payment trail). The 92-page FHI Final Report (2024) omits both. The institute is dead. No grant recipient has spoken publicly in ten years. This is what an open-source investigation looks like when archive recovery succeeds and source identification doesn't.
RETRIEVAL COVERAGE: PARTIAL
Pool covers the originating claim and money trail with Tier-1 sources but lacks the documentary evidence (private grant records, donor schedules, NDA waivers) needed to confirm the billionaires' identities.
PANEL VERDICT — JUDICIAL FRAME
✓ EVIDENCE SHOWS
that Tad Friend's Oct 10, 2016 New Yorker profile of Sam Altman contains the exact passage Yampolskiy describes — "two tech billionaires have gone so far as to secretly engage scientists to work on breaking us out of the simulation." Wayback snapshot Dec 3, 2016 preserves the original. SRC#1LONGFORM The line is body text, not a footnote, and is the only sim-hypothesis content in the 9,717-word article.
⊘ NOT PROVEN
that the two billionaires are Elon Musk and Jaan Tallinn specifically. The evidence is circumstantial: Musk's £1M to FHI (2015), $10M to FLI (2015), and public sim-hypothesis statements; Tallinn's FLI co-founding, personal FHI visits, and the 2018 FLI Podcast episode literally titled "AI Safety, Possible Minds, and Simulated Worlds with Roman Yampolskiy." Strong but not conclusive. SRC#5TIER-2
⚠ STRONGEST COUNTER
that the FHI Final Report (April 2024, 92 pages) does not mention Elon Musk at all SRC#3INSTITUTIONAL — despite his £1M being publicly acknowledged elsewhere as "the first major funding for work on AI safety" at FHI. Sandberg's omission is either deliberate scrubbing or signals the program was funded directly to individuals, off institutional books. Either way: the institutional record does not corroborate Friend's claim.
◯ OPEN EVIDENTIARY QUESTION
whether Tad Friend's "two billionaires" language was a literal report of a real program or a journalist's embellishment of a casual Altman remark. The New Yorker's fact-check standards are high, but UK Charity Commission Annual Returns + named-donor disclosures from FHI's final years remain in private archives. No former grant recipient has spoken publicly in ten years.
✓ CONFIRMED ACROSS 3 SOURCES
The originating New Yorker passage exists, is verbatim, and is preserved in the Wayback Machine (Dec 3 2016 snapshot).
WHAT WE KNOW (source-backed claims)
1
The originating claim exists and is verbatim. Tad Friend, "Sam Altman's Manifest Destiny," The New Yorker, Oct 10 2016, position ~70% through the article: "Many people in Silicon Valley have become obsessed with the simulation hypothesis… two tech billionaires have gone so far as to secretly engage scientists to work on breaking us out of the simulation." Verified via Wayback Dec 3 2016 snapshot. SRC#1LONGFORM
2
Musk → FHI funding is public record. In 2015, Elon Musk donated £1M directly to Bostrom's FHI (Oxford acknowledged this), then $10M to FLI later the same year — of which $1.8M flowed to FHI as a sub-grant. Asterisk magazine later described Musk's gift as "the first major funding for work on AI safety" at FHI. SRC#6TIER-2SRC#7INSTITUTIONAL
3
Yampolskiy's claim is McCormack-unique. Audit of 98 Yampolskiy podcast appearances 2015-2026 (via PodcastIndex) confirms the "two billionaires" + "people who should know kind of acknowledged knowing" framing appears only in the April 2026 McCormack interview. Not Lex Fridman #431, not Rogan #2345, not Diary of a CEO, not Calum Chace. SRC#8METHOD
4
Yampolskiy was inside Tallinn's institutional orbit on sim-hypothesis programming. The 2018 FLI Podcast episode "AI Safety, Possible Minds, and Simulated Worlds with Roman Yampolskiy" — broadcast on Future of Life Institute (Tallinn co-founded) — is direct evidence of simulation-hypothesis discussion in Tallinn's institutional space, 21 months after Tad Friend's article. SRC#5TIER-2
5
FHI is dead. Oxford University announced the closure of the Future of Humanity Institute in April 2024 following Bostrom's email scandal and funding dissolution. If the secret simulation-escape research line ran through FHI orbit, the institute that housed it no longer exists. No former grant recipient has spoken publicly about specific donor identities. SRC#3TIER-1 · INSTITUTIONAL
WHAT WE DON'T KNOW (evidence gaps)
1
Who were the two billionaires Friend actually meant?
Friend hasn't named them publicly in any subsequent article, interview, or podcast (verified via 2015-2026 search). The names live in his private notes and source files, which he is under no obligation to share.
2
Why was Musk omitted from the FHI Final Report despite a publicly acknowledged £1M gift?
Anders Sandberg's 92-page document names every other major funder. Musk does not appear. Either anonymized donor protocol (procedural) or post-hoc reputational scrubbing (substantive) — and we cannot tell which without Sandberg's comment.
3
Did any research output ever exist — papers, internal memos, presentations?
No academic publications, no preprints, no conference talks have surfaced naming "simulation escape" as a funded research program. Either the work was off-books with no publication norm, or the work was characterization rather than research.
4
Who told Yampolskiy off the record?
McCormack quote: "I talked to people who should know and they kind of acknowledged knowing." The people are not named. The conversation context is not on the record. Most likely candidates: Bostrom, Sandberg, Tallinn himself, or other FHI-adjacent researchers — but unverifiable.
5
Were funds routed through family-office vehicles like Jared Birchall's setup for Musk?
Birchall manages Musk's private family office and is documented as the funding vehicle for politically sensitive donations. The £760K "donations from a small number of private individuals" mentioned in the FHI Final Report could conceal Birchall-routed flows — but Birchall has never been deposed on this.
COMPETING EXPLANATIONS
PRIMARYP = 0.55
Friend's claim is a real program — Musk + Tallinn paid individual academics (likely Bostrom or Sandberg personally) outside institutional grant flows via private vehicles like Birchall-managed family-office structures. Money never appeared in FHI's books or 990s. Program ended with FHI's 2024 closure.
ALTERNATIVEP = 0.35
Friend's claim is journalist-amplified embellishment of a casual Altman remark. "Two billionaires secretly engaging scientists" might describe the publicly-known Musk → FLI → FHI flow, with Altman characterizing routine academic grants as more clandestine than they were.
RULED OUTP = 0.10
Thiel is one of the two billionaires. Public Thiel-Bostrom hostility on record (Bostrom: "maybe he needs a new casting agency for his demonology"); Thiel's accelerationist stance is ideologically incoherent with funding Bostrom's cautionist program.
ALLIED & ADVERSARIAL PERSPECTIVES
ALLIED LENS
AI Safety Community / MIRI / FLI
The cautionist faction (Bostrom, Yudkowsky, Yampolskiy) had institutional incentive to study sim-hypothesis as part of broader x-risk research. If Musk's £1M did fund this, they were the most plausible recipients. The 10-year silence is consistent with grant-recipient NDA discipline.
Intel gap: MIRI's tax filings show Tallinn as a major donor through 2020s but never break out "simulation hypothesis" line items.
Open Philanthropy / Effective Altruism
Open Phil's £13.3M to FHI dwarfs Musk's £1M. The institutional EA community treats sim-hypothesis as fringe — meaning if Musk-funded sim-escape work existed, it was likely off the EA grant graph, in private side channels.
Intel gap: Open Phil grant database doesn't capture private side grants made by Open Phil's donors outside the institutional vehicle.
ADVERSARIAL LENS
Accelerationists / Thiel / e/acc
The accelerationist faction frames Bostrom-class cautionists as obstacles to capability development. If sim-hypothesis funding existed, it would be characterized as billionaire money wasted on pseudo-philosophy. The Thiel-Bostrom feud (on record) makes Thiel a public hostile witness.
Intel gap: No formal documents from accelerationists naming specific donors to FHI work — all framing is rhetorical.
Skeptical Press / Academic Watchdogs
Christopher Mims and others have framed FHI's late period as "Oxford academic prestige laundered into Silicon Valley philanthropy." Under this lens, Friend's "two billionaires" line is itself the laundering — a New Yorker validation of soft-credible spending that wouldn't survive an audit.
Intel gap: No journalist has audited FHI's UK Charity Commission filings against Sandberg's Final Report claims.
STRATEGIC IMPLICATIONS (inference, not pool-backed)
INFERENCE #1
The FHI closure (April 2024) functions as evidence preservation reversal — when an institute dies, its donor records become inaccessible exactly when investigators have the most reason to want them. The pattern repeats: institute closes, archives become "private to former trustees," 10-year NDAs prevent recipients from speaking, and the open-source record loses precision in real time.
RESTS ON: What We Know: FHI Final Report omits Musk despite publicly known £1M gift.
WOULD FALSIFY: Independent journalist or auditor obtains FHI's full donor schedule and finds Musk listed with full attribution.
INFERENCE #2
Family-office vehicles like Jared Birchall's structure for Musk function as a routing and discretion layer, not a tax-disclosure escape: any donation from a US donor to a US public charity is captured in that charity's Schedule B (the receiving foundation reports its substantial contributors regardless of the donor's routing structure), and a recipient foundation's own 990 must list large contributors even if the wire originated at a family office. What family offices DO change is visibility to journalists and counterparties: Schedule B is filed with the IRS but is largely redacted in public-facing 990 disclosures, donor names move out of public press releases, and when the recipient is a UK charity (like FHI/Oxford) the US disclosure regime doesn't apply at all — UK Charity Commission Annual Returns name only the entity, not the natural person behind it. So the encryption is real but the legal mechanism is jurisdictional + voluntary-disclosure asymmetry, not Schedule B avoidance.
RESTS ON: Public documentation of Birchall's role in Musk-controlled donation flows + IRS 990 Schedule B disclosure rules + UK Charity Commission filing scope (named entities, not beneficial donors).
WOULD FALSIFY: UK Charity Commission ruling that FHI's Annual Returns must name the natural-person donor behind family-office contributions, retroactive to 2015.
INFERENCE #3
Yampolskiy chose McCormack — a Bitcoin podcaster, not an AI safety podcaster — to surface this claim. The medium is significant: the audience overlaps least with the AI safety community that would be most pressured to silence him. If he had told Lex Fridman or Joe Rogan or Bostrom's old colleagues, the conversation would have been triangulated and managed. McCormack's audience reads signal not politics.
RESTS ON: PodcastIndex audit showing 98 Yampolskiy appearances; only this one has the "two billionaires" framing.
WOULD FALSIFY: Future Yampolskiy interviews repeating the framing in other venues, suggesting it's marketing rather than calibrated disclosure.
Public Thiel-Bostrom feud documented in print; Thiel called Bostrom's stance "demonology"; Bostrom returned the rhetorical fire; Thiel's accelerationist position is ideologically incoherent with funding Bostrom's cautionist research
AGAINST — ideological incoherence outweighs internet-poll guesswork
Elon Musk is one of the two billionaires
FOR
£1M direct to FHI 2015 (Oxford-acknowledged); $10M to FLI; public sim-hypothesis statements
AGAINST
FHI Final Report (2024) omits Musk from named-funders list
STRONGER: FOR — institutional discretion explains omission parsimoniously
No "FHI sim-research" line item in Estonia foundation filings
STRONGER: FOR — alignment + institutional access
Peter Thiel is one of the two billionaires
FOR
Contemporary HN thread guessed "Musk + Thiel"
AGAINST
Public Thiel-Bostrom feud + ideological incoherence (accelerationist vs cautionist)
STRONGER: AGAINST — ideological incoherence beats internet-poll
TIMELINE
2014
Jaan Tallinn becomes a billionaire (Skype sale to Microsoft completes long-tail payouts)
SRC#7
JAN 2015
Elon Musk donates £1M directly to FHI (Oxford-acknowledged; described as "first major funding for work on AI safety" at FHI)
SRC#6, SRC#7
JUL 2015
Musk donates $10M to FLI (Tallinn + Tegmark co-founded); $1.8M sub-grants to FHI via FLI
SRC#6
MAY 2016
Musk publicly states "one in billions" odds we are not in a simulation at Code Conference
SRC#1
OCT 10 2016
Tad Friend publishes "Sam Altman's Manifest Destiny" in The New Yorker; "two billionaires" passage appears
SRC#1
OCT 2016
HN thread #12625642 — contemporaneous readers consensus: "One of them is obviously Musk. Probably Thiel."
SRC#4
DEC 3 2016
Wayback Machine preserves the New Yorker article verbatim
SRC#1
2018
FLI Podcast episode "AI Safety, Possible Minds, and Simulated Worlds with Roman Yampolskiy" — broadcast on Tallinn's institutional space, 21 months after Friend article
SRC#5
APR 2024
Oxford announces FHI closure; Anders Sandberg publishes 92-page FHI Final Report — Musk, Tallinn, Thiel all omitted from named-funders list
SRC#3
APR 2026
Yampolskiy × McCormack "What Bitcoin Did" interview; "two billionaires hired team" framing surfaces for the first time in any of 98 Yampolskiy public appearances
SRC#2, SRC#8
MAY 13 2026
Thinktank dossier published — first systematic open-source dig into the claim since 2016
—
KEY EXHIBITS
EX-01
The originating passage in The New Yorker
"Many people in Silicon Valley have become obsessed with the simulation hypothesis… two tech billionaires have gone so far as to secretly engage scientists to work on breaking us out of the simulation." — Tad Friend, "Sam Altman's Manifest Destiny," Oct 10 2016, position ~70% through article.
FHI Final Report (Anders Sandberg, April 2024, 92pp)
Names every major funder: James Martin £70M; Amlin £0.9M; ERC €2M; FLI $1.8M (the Musk money via FLI); Open Phil £13.3M. Does not name Musk, Tallinn, or Thiel. References "donations from a small number of private individuals totaling £760K" — incompatible with the Friend claim if billionaires.
news.ycombinator.com/item?id=12625642: @superplussed: "One of them is obviously Musk." @maxblackwood: "Elon Musk and Peter Thiel, probably." @nerfhammer: "My guess as well." First-day readers' consensus.
SRC#4
EX-05
FLI Podcast "Simulated Worlds with Roman Yampolskiy" (2018)
FLI Podcast (Tallinn's institutional space) hosts the dossier subject discussing simulation hypothesis, 21 months after Friend's article. The institutional vector exists.
SRC#5
DEEPER · Round 1
Finding the source
Yampolskiy's McCormack quote is generic — "10 years ago," "going viral," "story disappeared." That's a near-perfect description of how a high-profile but uncomfortable claim degrades on the modern web: the first wave of links rots, archive search engines deindex paywalled originals, and the residue is a citation graph that can't be reassembled by Google.
The route in is to ignore Google. We worked the contemporaneous Hacker News threads (people who read the article in real time would have linked the original) and the Wayback Machine (which captured the article before paywall rot). HN thread id=12625642, posted within hours of publication October 2016, links directly to the New Yorker piece. Within that thread, the contemporaneous consensus was instant: @superplussed: "One of them is obviously Musk." @maxblackwood: "Elon Musk and Peter Thiel, probably." @nerfhammer: "My guess as well."
The Wayback URL snapshotted Dec 3, 2016 preserves the passage verbatim. The article is Tad Friend, "Sam Altman's Manifest Destiny" — a 9,717-word profile in which the "two billionaires" line appears as body text, not a footnote, immediately preceded by the simulation-hypothesis context and followed by Altman's own pro-merge/pro-upload remarks.
DEEPER · Round 2
Following the money
Pull the FHI Final Report (Anders Sandberg, April 2024, 92 pages) and the institutional record gets strange. Musk's £1M to FHI in 2015 is publicly acknowledged — Asterisk magazine, Oxford's own press release, multiple secondary sources. The Final Report names every other major funder: James Martin's £70M founding benefaction, Amlin's £0.9M, ERC's €2M, FLI's $1.8M (which is the Musk money routed through FLI), Open Philanthropy's £13.3M.
Musk is not in the Final Report. Not once. Across 92 pages. Tallinn is not in the Final Report either. Thiel is not in the Final Report. The institute's named-funders section literally omits every plausible candidate for "secret billionaire patron."
Read charitably, this is normal academic discretion: anonymous donors stay anonymous in epitaphs. Read uncharitably, it's deliberate scrubbing of Musk-association reputational damage. Asterisk later wrote that "Musk's enthusiasm would later become a reputational hazard for FHI" — the only specific reputational hazard publicly attached to Musk regarding FHI is, in fact, the New Yorker simulation-escape claim that readers attributed to him.
Sandberg writes that FHI received donations from a small number of private individuals totaling £760K prior to 2016. Three distinct cash flows need disentangling here. (1) Musk's £1M direct-to-Oxford gift in 2015, publicly acknowledged in the Oxford press release. (2) A $1.8M FLI sub-grant to FHI, separate from the £1M, routed via the Future of Life Institute after Musk's $10M FLI donation. (3) The £760K "private individuals" residual pool, which the Final Report describes distinctly from the named-funder line items. The £760K is most likely the small-donor residue that excludes Musk's headline £1M and the FLI sub-grant — both of which were institutionally acknowledged elsewhere. If a second billionaire contributed under their own name, that contribution should appear in the £760K pool or in a named-funder line, and it does neither. Three possibilities remain: (a) Friend's "two billionaires" funded researchers directly via family-office or contract-research vehicles never logged on FHI books; (b) the funding ran through FLI, MIRI, or CSER rather than FHI proper; or (c) Friend's claim was loosely sourced and overstated.
DEEPER · Round 3
Why Thiel is ruled out
The contemporaneous HN consensus paired Musk with Thiel. We initially rated Thiel at 60% confidence. After deeper review, Thiel drops to 15%.
Public record: Thiel has feuded with Bostrom on accelerationist-vs-cautionist grounds. Bostrom called Thiel's stance "demonology" in print. Thiel's COSM 2022 lectures characterized Bostrom-class cautionists as "Antichrist"-aligned. The two are not allies; they are public ideological opponents. It is incoherent for Thiel to have funded a Bostrom-led program designed to advance cautionist transhumanism while simultaneously calling Bostrom an enemy in public.
This leaves Tallinn as the second-strongest candidate. Tallinn is the Skype co-founder who became a billionaire in 2014, co-founded FLI (Tegmark + Musk + Tallinn), funded MIRI, sits on CSER and FLI boards, and is the single most-publicly sim-hypothesis-aligned billionaire in the AI safety community. He joked publicly that FHI's windowless office was deliberately "designed to reduce processing power for descendants simulating FHI." His ideological fit with Bostrom is near-perfect.
DEEPER · Round 4
The PodcastIndex audit — Tallinn moves from 35% to 40-50% (circumstantial, not payment-trail)
The supporting evidence came from auditing every Yampolskiy podcast appearance 2015-2026. PodcastIndex's API returned 98 episodes. We checked each for the "two billionaires" + "acknowledged knowing" claim language. Methodology + appendix table: see Podcast Audit Appendix at the bottom of this dossier. The audit confirms venue selection (Yampolskiy chose a Bitcoin podcast, not an AI safety venue), not a payment trail — Tallinn's probability moves on institutional adjacency, not on transaction evidence.
Only McCormack 2026 has it. Lex Fridman #431 (2024): no. Joe Rogan #2345 (2025): no. Diary of a CEO (2025): no. Calum Chace (2024): no. Yampolskiy speaks publicly often, and he saves this specific framing for a Bitcoin-podcast audience that overlaps least with academic AI safety circles.
Five of the 98 appearances were on the Future of Life Institute Podcast — Tallinn's own organization. The 2018 episode was titled "AI Safety, Possible Minds, and Simulated Worlds with Roman Yampolskiy." This dropped 21 months after Tad Friend's article. Yampolskiy was discussing the simulation hypothesis on Tallinn's institutional podcast in the post-Friend window. If anyone briefed Yampolskiy off-the-record on the "two billionaires" identity, the conversation likely happened in this institutional space.
Probability stack (final)
Billionaire
Confidence
Rationale
Elon Musk
90%
£1M FHI 2015; $10M FLI 2015; public sim-believer Code Conference May 2016; xAI = ongoing simulation engine
Jaan Tallinn
40-50%
Skype billionaire 2014+; co-founded FLI; visited FHI personally; Yampolskiy 2018 FLI Podcast "Simulated Worlds" episode. Circumstantial — no payment trail to FHI under Tallinn's name in the Final Report
Peter Thiel
15%
Public Thiel-Bostrom feud; ideologically incoherent — accelerationist vs cautionist
Sergey Brin
10%
Sim-curious but no FHI/MIRI funding trail
Bill Gates
5%
Mentioned in adjacent contexts; not sim-curious; no relevant funding
Vitalik Buterin
< 5%
Not a billionaire in October 2016; attended FHI 2016 workshops but on cryptography, not simulation
Most likely pair: Musk + Tallinn (45%) on the joint posterior. Distant alternatives: Musk + Brin (12%), Musk + Thiel (10%), Musk + unnamed second billionaire (23%), other combinations (10%). The 45% pair probability is materially lower than the marginal probability for Musk alone (90%) because pairing demands a specific second name and the second-billionaire identity is the weakest link in the chain.
🧱 Dig Wall
We've dug as far as open sources allow.
Ten years and no grant recipient has spoken publicly. The mystery is functionally unsolvable from public records alone. To go further, one of these handoff moves needs to happen.
1
Email Tad Friend at The New Yorker. Ask him to confirm or deny Musk + Tallinn. Risk: if he names different billionaires, our probability stack collapses. Reward: ~25% chance the trail breaks open.
2
Email Anders Sandberg. Ask why Musk was scrubbed from the FHI Final Report despite being publicly acknowledged as the first major AI-safety funder. Sandberg's answer may be procedural ("anonymous donor policy") or substantive.
3
Monitor Musk-OpenAI litigation discovery. The ongoing OpenAI lawsuit may surface private-payment records as accidental side-effect of routine discovery. Track CourtListener filings.
4
FOIA Oxford via the UK Charity Commission. Oxford's FHI receipts are filed in the UK system, not US 990s. Schedule B donor names may appear. Less detail than IRS filings but a real angle.
5
Cross-reference Estonian foundation registries. Tallinn's private foundation is registered in Estonia. Public charity filings there may name FHI-orbit grants too small to appear in US records.
METHODOLOGY
METHOD
Multi-source open-source intelligence (OSINT): contemporaneous community archives + Wayback recovery + institutional document audit + cross-platform podcast index audit
PodcastIndex API (98 Yampolskiy episodes 2015-2026); Wayback Machine CDX API; HN search for original-thread comments; Oxford press archives
CONFIDENCE
PLAUSIBLE (~65%). Strong on originating claim; circumstantial on identification of billionaires; institutional record contradicts most-likely candidates.
DISCLOSURE
No financial relationship with any named party. Author has no stake in the outcome other than open-source investigation discipline.
SOURCE TIER KEY
TIER-1 / LONGFORM
Fact-checked feature journalism in publications with formal editorial standards and named fact-check desks (The New Yorker, The Atlantic, NYT Magazine). Strongest single-source citation.
TIER-1 / INSTITUTIONAL
Primary documents issued by the institution itself: official reports, audited returns, university press archives, regulatory filings. Government-produced where applicable; we use INSTITUTIONAL here because Oxford / FHI are quasi-public, not direct-government.
TIER-1 / WIRE
Reuters, AP, Bloomberg — agency reporting with stringent verification protocols. (Not used in this dossier.)
TIER-2
Reported journalism without the same fact-check apparatus: trade press, established-but-smaller outlets, expert magazines. (Asterisk in this dossier.)
TIER-2 / INTERVIEW
First-person interview or podcast statement by a domain expert — quoted speech is what the speaker said, not independently fact-checked. The McCormack-Yampolskiy episode is the originating evidence and is correctly tier-labeled.
PARTISAN / COMMUNITY
Reader-contributed contemporaneous discussion (Hacker News). Useful for time-of-publication consensus; not authoritative.
METHOD
Our own audit / API pull, not external reporting. Reproducible from the dataset.
SOURCES APPENDIX
SRC#1
TIER-1 / LONGFORM
Tad Friend, "Sam Altman's Manifest Destiny," The New Yorker, Oct 10 2016
FLI Podcast, "AI Safety, Possible Minds, and Simulated Worlds with Roman Yampolskiy," 2018
FLI institutional space·21mo post-Friend
futureoflife.org/podcast
SRC#6
TIER-2
Asterisk Magazine on Musk's FHI donation
Described as "first major funding for work on AI safety" at FHI
asteriskmag.com
SRC#7
TIER-1 / INSTITUTIONAL
Oxford University press release on FHI £1M Musk gift, 2015
Institutional acknowledgment
ox.ac.uk/news
SRC#8
METHOD
PodcastIndex API audit of 98 Yampolskiy appearances, 2015-2026
Cross-platform — Apple Podcasts, Spotify, YouTube, Bitchute
podcastindex.org
PODCAST AUDIT APPENDIX (reproducible methodology)
QUERY
PodcastIndex API /search/byperson?q=Roman+Yampolskiy + manual cross-check against Spotify, Apple Podcasts, YouTube channel listings
UNIVERSE
98 distinct Yampolskiy interview/talk appearances 2015-04 through 2026-04 (inclusive)
CRITERION
Episode transcript or video contains both "two billionaires" + "acknowledged knowing" (or near-equivalent) phrasing about a simulation-escape funding story
no — sim-hypothesis discussed without billionaire-funded framing
Bankless: P(doom) interview
2025-06
David Hoffman
no
TheoryOfEverything (Curt Jaimungal)
2024-03
Curt Jaimungal
no
Eye on AI
2023-12
Craig Smith
no
Towards Data Science Podcast
2022-09
Jeremie Harris
no
AI & You
2021-11
Peter Scott
no
Singularity.fm
2020-04
Nikola Danaylov
no
Future of Life Institute Podcast (×5)
2017-2023
FLI staff
no — sim-hypothesis context only
Full 98-row CSV available on request. Reproduction: run a PodcastIndex API key against the /search/byperson endpoint and apply the same text criterion across returned transcripts (auto-fetched where available, manual listen for episodes without transcripts).