Diane Francis

Who she is

Affiliation: Editor-at-Large, National Post/Postmedia (since 1998) — role check 2026-09-21: LinkedIn headline begins "Weekly columnist National Post and Kyiv…"; Kyiv Post author page shows her 09-06 and 09-15 columns; nationalpost.com would not fetch, so the NP byline rests on mastheads and reprints. Columnist, Kyiv Post (ten bylines 2026-06-25 → 09-15). Nonresident Senior Fellow, Atlantic Council Eurasia Center — listed; newest AC piece 2024-01-18; nominal. Paid twice-weekly Substack ("Power, money, tech, corruption, geopolitics"; "readers in 106 countries"). Former Singularity University adjunct faculty — absent from SU's experts directory (checked 2026-09-21) though her bios still say "Faculty." Member, Abundance360 (Diamandis's paid community) and The Progress Network. Bio boards: Hudson Kleptocracy Initiative, Canada-US Law Institute. Born Chicago 1946-11-14 (79); first woman editor of a Canadian national daily (Financial Post, 1991–98); ten books, 1986–2013, none since.
One-line position: Great-power politics frames everything — Putin's war, Trump's tariffs, and AI as the next nuclear weapon: "AI will be the defining strategic, economic, and geopolitical prize of the 21st century, and unless bridled, will also outsmart humanity" (World War AI, 2026-07-06). A relay, not a source: she forwards Hinton, Amodei, Yudkowsky and Zelensky to a Canadian and Ukraine-diaspora readership, with a hawk's valence (lean) and loose numbers.

Discipline & technical bet

Investigative business journalist (Bre-X, "who owns Canada," kleptocracy — rule by thieves) turned geopolitics columnist. No technical bet of her own: she does not build, fund or test AI. The v1 "AI investor" line has no source — her own site lists two gold-mining directorships (Aurizon, Lake Shore Gold; both sold) and unnamed software/medical-device involvements. Her wager is editorial: AI belongs on the geopolitics page — weapons, model-safety events and market concentration are one story about power. The technology bet under her writing is Ukraine's: home-grown drones and Sapsan missiles will decide the war. Technology-beats-mass, on a battlefield.

Key claims (Says)

• Empirical (v1): AI/AGI are strategic assets in great-power competition. Held — World War AI (2026-07-06) rests on Zelensky's UN line, "the most destructive arms race in human history, because this time it includes artificial intelligence" (2025-09-24), and Anthropic's June-2026 self-improvement report.
• Empirical (v1): geopolitical framing — US-China, NATO, Russia. Held; China barely appears in her 2026 output — her great-power story is Russia.
• Empirical (v1): Silicon Valley as a beat tied to energy, white-collar crime and corruption. Held — 2026 columns link AI to market "circularity" (2026-08-06), to publicly funded research and a sovereign wealth fund (a government-owned investment pot; 2026-06-08), and to public safety (2026-08-24).
• Normative: governments must "bridle" AI; "Safety and damage control should be front and center" (2026-04-27); Sanders' 50%-of-equity public fund is "only fair" because Silicon Valley "was largely built on fundamental, publicly funded research" (2026-06-08). Held as a position since 2023-04-06; a value, not graded for truth.
• Empirical: "on April 7, a monumental tipping point in technology was reached" — Anthropic withheld Claude Mythos Preview. Held on the facts — Project Glasswing (2026-04-07) names the model and partners; extended to critical infrastructure in 15+ countries (TechCrunch, 2026-06-02). "Tipping point" is interpretation.
• Empirical: "OpenAI, run by Sam Altman, grabbed the contract and took on the Pentagon's plans to create autonomous lethal weapons and spy on Americans" (2026-04-27). Wrong on mechanism — OpenAI's published terms prohibit mass domestic surveillance, autonomous weapon systems and high-stakes automated decisions, cloud-only with cleared staff in the loop (TechCrunch, 2026-03-01, quoting OpenAI; OpenAI's own post 403 on fetch). Valence partly live: the contract is unpublished and the Pentagon later asked for "minimal refusal rates" (The Intercept, 2026-09-08) — the loophole worry is fair, the flat claim is not. Her line that Amodei refused Pentagon use of Claude for fully autonomous weapons Held (same source).
• Empirical (v1): "Techno-optimism, qualified — insider who maintains journalistic distance." Drifted — distance yes, optimism no: every AI column 2023→2026 is alarm-first (detail under Theories).
• Empirical: the AI rally is "a selective bubble inside an overvalued market," not a 2000 rerun — the giants "generate more cash flows than most nation-states"; still, "putting all your savings into the Magnificent Seven would be dangerously concentrated" (2026-08-06). Live — check below. She credits the verdict to ChatGPT; her one bullish AI claim is outsourced.
• Empirical: chatbots are implicated in killings and no AI company has "a foolproof, preventative process to identify, impede, or report threats" (2026-08-24). Held on the case facts — OpenAI banned the Tumbler Ridge shooter in June 2025 and did not notify police (CBC; NPR, 2026-04-29). Drifted on mechanism — she writes OpenAI "allowed her back in"; OpenAI says she evaded the ban with a second account found only after police named her (Reuters via Yahoo). Banned-and-evaded is not reinstated.
• Empirical / normative: Russia is already at war with Europe by hybrid means (attacks kept below open war); "Europe's weakness is not military. It is psychological" (2026-09-21). Recommends banning all Russians and their firms, seizing the shadow fleet, confiscating assets. Check below.
• Empirical: Russia is losing — 1.47M casualties (Ukrainian General Staff-derived, unverified here), record gold sales, elite dissent (2026-08-20). The gold sale is real: 44 t in H1 2026, largest in 25+ years, ~US$5.6B (Kitco/Reuters, 2026-07-21). Her sentence: "Russia had sold 44 million metric tonnes of gold since January 2026, worth $6.2 billion and equivalent to half of its total gold reserves." Wrong on the numbers — 44 t, not 44 million (all gold ever mined ≈ 216,000 t), and ~2% of Russia's 2,282 t, not half. Attributed to Poland's foreign minister; printed uncorrected.
• Unfalsifiable — rewrite or retire (not graded): "unless bridled, will also outsmart humanity" (2026-07-06); AGI "will upend human existence" (2025-10-30); "AI imperils the world order and humanity itself" (2026-04-27); firms flee offshore if a Sanders-style levy advances (2026-06-08). No observable, no date; the rewritable ones are rewritten below.

Notable predictions — with falsifiable checks

• (2023-04-06) AI will replace writers, mathematicians, artists, doctors, teachers and engineers "within years." Wrong at market level (2026-09) — Yale Budget Lab's CPS tracking finds no discernible economy-wide AI footprint through its 2026 updates; Stanford Canaries (data to 2026-06): "no widespread displacement," but 22–25-year-olds in AI-exposed jobs are 19% below trend. Right at the entry margin; wrong on occupations and scale. Re-grade: BLS physicians, K-12 teachers or engineers down ≥10% from 2023 by 2028-04.
• (2025-10-30) AGI is "only a few years away." Live. Check 2028-10-30: a frontier lab or recognized arbiter (e.g., a resolved Metaculus AGI question) declares AGI → Held; else Wrong.
• (2026-04-27, relaying Amodei, 2025-05) "AI could eliminate 50% of all entry-level white collar jobs in five years." Live. Check 2030-05-28: entry-level (22–25) office employment ≥50% below 2025 → Held; interim 2027-06, Canaries gap stalled below 25% → Drifted.
• (2026-08-06) a 20–30% correction is plausible. Live. Check 2027-08-06: Nasdaq-100 ≥20% below its 2026 peak → Held; never past 10% → Wrong; between → Drifted.
• (2026-06-08) firms relocate offshore if a Sanders-style levy advances; cites Thiel's Argentina move — that fact Held (NYT-reported; US$12M Buenos Aires home; "temporary"; 2026-05-30). Rewritten check: if an AI equity/sovereign-fund bill passes either chamber by 2027-12-31, count Mag-7 or frontier-lab HQ moves within 12 months.
• (2026-02-05) "What is underway with Russia is not a peace process but a slow-motion humanitarian catastrophe"; Putin keeps fighting absent decisive Western action. Held — ISW's assessment of 2026-09-20 reports continued Russian offensive operations; her Kyiv Post columns of 09-06 and 09-15 describe a live war.
• (2026-02-05) if Trump loses Congress in the midterms, his generous terms to Putin end. Live. Check: 2026-11-03 result; a Democratic chamber plus documented hardening (sanctions/aid) by 2027-03-01 → Held; no change → Wrong.
• (2026-05-21 → 06-29 → 07-09) Ukrainian ballistic missiles are "soon," "in a few months," "in the coming weeks," and "will change everything in this war." Drifted — Zelensky confirmed combat use of the Sapsan (300–500 km; theater-range, not intercontinental) on 2025-12-09, six months before "ready in a few months" (Euromaidan Press); her line about joining "the elite group of nations with the technology ... to operate intercontinental ballistic missiles" has no source. Game-changer half live: check 2026-12-31 for an ISW-documented Russian defensive shift or Kremlin territorial concessions.
• (2026-08-20) "Then, hopefully by year-end, Ukraine will have ballistic missiles and have reoccupied Crimea. That is when twilight ends for Putin." Live — the boldest call on the file. Check 2026-12-31: Ukraine controls Crimea → Held; no change → Wrong ("hopefully" does not rescue a dated claim). Base rate: Russian decline has been her forecast since 2022; no dated version has landed.
• (2026-09-21) Europe is already at war with Russia and must respond "aggressively." Check 2027-09-21: a NATO member treats a hybrid incident as an armed attack (Article 4/5) or the EU adopts a blanket Russian-entry ban → Held; else rhetorical → Drifted.
• (2013, Merger of the Century) Canada and America should become one country; Canada must build strength or lose leverage. Signature-framing check: Drifted — she now calls the "51st state" version "totally off-base" and backs an EU-style union that keeps sovereignty (2024-12-30), while claiming vindication on dependence: "Nothing changed" (2026-09-14). Mechanism (dependence invites coercion) Held — Feb-2025 tariffs and the 51st-state campaign (Fortune, 2025-01-08; CBC); the remedy (merger) retired by its author.

Revealed behavior (Does)

• Time: last twelve Substack posts (2026-08-13 → 09-21): four Ukraine/Russia, two Canada, two Trump, two AI, two energy/other. AI is one column in six; Ukraine is the beat — v1's "AI, technology, geopolitics" order is backwards.
• Money: the paid Substack (two paid issues a week, four weeks off) is the business; no AI investment, advisory or board role found; past directorships were gold miners. She pays into Abundance360 — Diamandis's customer, not partner.
• Distribution: one column, four mastheads — Substack, National Post, Kyiv Post, Ukrainian World Congress reprints.
• Sourcing: AI columns are built from others' announcements and quotes (Hinton, Amodei, Yudkowsky, Zelensky, Sanders, Sacks, Thiel); no researcher interviews, no papers. Ukraine columns lean on Ukrainian officials ("explained an official").
• Fact discipline: accurate on events (Mythos/Glasswing, the June-2026 report, Thiel, Moreno-Gama, Tumbler Ridge, Zelensky at the UN); loose on specifics (Pentagon terms, gold tonnage, "intercontinental," "allowed her back in").
• Self-presentation lags: bios still list Ryerson (left 2018; renamed TMU 2022), SU faculty, Atlantic Council columns (last 2024-01). Her own bio is the least current source on her.

Feels

Hope that Ukrainian ingenuity beats Russian mass — "It was David slaying Goliath because David was smarter." Fear that the same ingenuity, in AI, ends with humanity outsmarted. Wants Canada taken seriously and Putin finished before she turns 80.

Hears

Ukrainian officials and the diaspora, Anne Applebaum's circle (Applebaum blurbs her Substack), Anthropic's safety communications, Hinton and Yudkowsky for AI alarm, Abundance360 and The Progress Network for optimism. Almost nothing from AI builders, labor economists or China-watchers.

Sees

Thirty-plus years inside Canadian business and kleptocracy reporting: she reads a Russian gold sale or a Bombardier ultimatum as a corporate-control move, which most AI commentators cannot. She also sees Ukraine's weapons industry through direct sources, and Canadian elite opinion from inside the room.

Incentive map

Re-derived this cycle. Who pays: paid subscribers (Ukraine diaspora, Canadian nationalists, doom-curious generalists), Postmedia, Kyiv Post; the AC fellowship is dormant; Abundance360 is money out, not in. What she sells: a hawk's clarity — Putin is losing, AI is the next nuke, Canada must arm. What she cannot say: "Ukraine may not get Crimea back," "the Pentagon deal has real safeguards," "AI hasn't dented employment yet" — each breaks the frame readers pay for. A paying member of Diamandis's optimism community publishing AI-alarm columns: alarm sells. Discount: high on Ukraine timelines and Russia-collapse calls; moderate on AI (credible sources, added heat, dropped nuance); low on Canadian business facts. Rule 25: no AI-company equity, advisory or board tie — audience capture, not capital.

Theories aligned with

• Great-power AI competition (v1 "geopolitical lens") — Held; the journalist's edition of the Aschenbrenner/Schmidt frame. No theory page on the bench.
• AI alignment / x-risk — relay variant: forwarding Yudkowsky, Tegmark, Hinton and Anthropic since 2023-04, with no probability or timeline of her own.
• Entry-level work crisis — via Amodei's 50% claim; she has not engaged the Canaries data.
• Techno-optimism (v1 "qualified") — Drifted: alarm-first on AI; optimism survives for Ukraine's defense tech ("They have reinvented war and built Europe's largest military-industrial complex," 2026-05-21) and AI equities ("generate more cash flows than most nation-states," 2026-08-06). Technology wins wars and markets; it endangers people.

What she's reacting against

• Western euphemism about Russia — "hybrid attacks," "peace process" — and leaders who "debate who the culprit is."
• Trump's transactional handling of Putin (a bombing pause Putin agrees to, then reneges on) and of Canada (51st state, tariffs, the Bombardier ultimatum).
• Silicon Valley self-regulation, and Altman's OpenAI specifically, against "conscientious" Anthropic.
• Canadian complacency — "Nothing changed" since her 2013 warning.
• The "51st state" idea itself — from the author of Merger of the Century.
• Chatbot companies' threat-reporting gaps after Tumbler Ridge.

Where she overlaps / splits (with Rich)

• Schmidt (v1): same geopolitics-of-AI lane; splits on posture — he advises the state, she wants it to bridle the labs and take half their equity.
• Diamandis (v1): overlaps institutionally (ex-SU, Abundance360), not in valence — she sits in his tent and publishes the opposite conclusion. Moonshots independence rule: customer, not partner.
• Hinton, Yudkowsky, Tegmark: she is their mainstream relay; splits in carrying no p(doom) or timeline of her own.
• Amodei: among the friendliest press voices for Anthropic's safety posture; splits sharply with Altman, and overstates his Pentagon deal.
• Aschenbrenner: shares "AI is the strategic prize"; splits on the adversary — his China, her Russia.
• Brynjolfsson, Frey: splits on evidence — she relays the 50% claim without their payroll data; her own 2023 job call fails on it.
• Crawford (v1): loose overlap — investigative rather than structural power critique; no evidence she reads Crawford.
• Thiel: splits — his Argentina move is her oligarch-flight warning, not a model.
• With Rich: the state-level "handed over the keys" question — her answer is that governments grab them back (sovereign fund, bridling, Glasswing-style gating). Her Ukraine columns are a data set for technology-beats-mass; her AI columns show how alarm propagates through mainstream press, one quote at a time.

Track record

• 1991–98: first woman editor of a Canadian national daily (Financial Post); Bre-X: The Inside Story (1998) and Who Owns Canada Now (2008) anchor the white-collar-crime and ownership credentials.
• 2013 Merger of the Century — political merger retired by its author (2024-12); the dependence warning vindicated by the 2025 tariffs. Right on mechanism, wrong on remedy — and the remedy was the title.
• Ukraine: hawk since Maidan; 2019 Friend of Ukraine Tryzub award; right after 2022 that Ukraine would not fold; repeatedly early on Russian collapse — "twilight" for four years.
• 2023-04 AI job-replacement call — Wrong at market level through 2026-09 (Yale Budget Lab; Stanford Canaries). 2026 AI event reporting — right on events, wrong on specifics (see Does).
• Pattern: right on direction, loose on dates and numbers; strongest on Canada, weakest on AI specifics; her best-calibrated sentence is the one she outsourced to ChatGPT.

Empirical vs normative

• Empirical: AI is the strategic prize; Ukraine has technological superiority; Russia is under fiscal strain; the AI rally is concentrated and partly circular; chatbots are implicated in killings.
• Normative: bridle AI by law; move half of big-AI equity into a public fund; treat hybrid attacks as war; Canada should arm and join an EU-style union with the US.
• Conflict note (CONVENTIONS rule 25): audience capture, not capital; no AI-company money. Weight event reporting moderately, Russia timelines lightly, Canadian business claims normally.

Weak spots / open questions

• No original AI reporting — no builder interviews, no papers, no China; relay risk compounds others' errors and adds her own ("44 million metric tonnes," "intercontinental," "autonomous lethal weapons" — all 2026, all while the underlying event was real).
• Russia-collapse forecasting has never carried a date that landed; the 2026-12-31 Crimea check is the first hard one.
• Bios ~8 years stale (Ryerson, SU, Atlantic Council); age 79 — continuity of the Substack is open, hence the silence wake trigger.
• Unverified, not graded: "AI investor" (v1; no source); "Two assassination attempts have been made against OpenAI's Sam Altman" (one confirmed — Moreno-Gama, 2026-04-10); "$27 trillion, or 36%" added to US stocks since 2022-11 (her figure); "1.47 million" Russian casualties (Ukrainian General Staff-derived).
• The v1 "Why she matters" premise — a bridge between tech-insider AI discourse and geopolitics — is half true: she bridges AI-safety discourse to a geopolitical readership; the insider half (SU, "investor") does not hold.

Rich's take

• (your synthesis here)

Delta log

2026-09-21 — v1→v2 migration + validation (bulk run)

• Grades: 18 Held / 7 Drifted / 3 Wrong; 4 unfalsifiable; 7 live checks (2026-11-03, 2026-12-31 ×2, 2027-08-06, 2027-09-21, 2028-10-30, 2030-05-28).
• Wrong: 2023-04 "AI will replace doctors, teachers, engineers within years" (Yale Budget Lab; Stanford Canaries, Aug 2026) · OpenAI "took on the Pentagon's plans to create autonomous lethal weapons and spy on Americans" (TechCrunch, 2026-03-01, quoting OpenAI's prohibited uses) · "44 million metric tonnes of gold ... half of its total gold reserves" (Kitco, 2026-07-21: 44 t of 2,282 t).
• Drifted: "techno-optimism, qualified"; Tumbler Ridge mechanism (Reuters via Yahoo); ballistic missiles "coming" (Sapsan in combat since 2025-12-09, Euromaidan Press); the 2013 merger thesis (2024-12-30); Atlantic Council "contributor" (last piece 2024-01-18, expert page); Substack "covers AI" (2 of 12 posts, archive); "fluency."
• Held: strategic-asset and geopolitical framing (World War AI); Silicon Valley beat; bridling position; Mythos/Glasswing facts (Anthropic); Amodei's refusal; "not a peace process" (ISW, 2026-09-20); the Thiel, Moreno-Gama, Zelensky-UN and self-improvement-report relays; NP role, former SU faculty, born Chicago 1946, ten books; Schmidt and Diamandis pairings; great-power AI competition.
• Most surprising delta: filed as an AI/tech persona on a bio line, but AI is one column in six and her signature thesis has flipped — the author of Merger of the Century calls Trump's 51st-state idea "totally off-base"; her boldest live call is Ukraine retaking Crimea by 2026-12-31 (Twilight for Putin).
• Employer/role check (mandatory): no job change — still Editor-at-Large, National Post (LinkedIn; Kyiv Post bylines to 2026-09-15). v1 corrections: not current SU faculty (SU directory); "AI investor" unsupported (own bio); Abundance360 and Progress Network added; AC fellowship nominal. Incentive map re-derived: audience capture, no AI-company money.
• Tier: quarterly — three dated checks mature by 2027-01 (midterms; Crimea/ballistic) and she is 79; drop to semiannual after the January-2027 review if they resolve.
• Confidence unchanged at medium: sources are plentiful and dated, but she is a relay; the file's value is in the checks.
• Research note: 25 searches, ~35 fetches; the search budget ran out after the last verification pass with no gap affecting a grade. OpenAI's own post returned 403 — that Wrong grade rests on TechCrunch's direct quotation.

Sources

• Her columns, 2025–26 (Substack; reprinted in National Post / Kyiv Post): AI then AGI (2025-10-30) · Putin's Ploy (2026-02-05) · Putin's failure (2026-04-22) · Tech's Tipping Point (2026-04-27) · Ukraine's Turning Point (2026-05-21) · Bernie's AI Fix (2026-06-08) · Ukraine's Game-Changer (2026-06-29) · World War AI (2026-07-06) · Ukraine's Watershed (2026-07-09) · The AI Casino (2026-08-06) · Twilight for Putin (2026-08-20) · AI Murders (2026-08-24) · America's AI Backlash (2026-08-31, paywalled) · Canada Falters (2026-09-14) · It's War Already, Europe (2026-09-21)
• Earlier: AI and Frankenstein (2023-04-06) · The "51st State" Insult (2024-12-30) · Ukrainian World Congress reprint
• Self-description and affiliations: Substack About · archive (topic count) · dianefrancis.com · Atlantic Council · Kyiv Post author page · LinkedIn · SU experts directory (name absent) · SingularityHub bio ("Adjunct Faculty") · The Progress Network · Wikipedia (birth date, book list, Ryerson to 2018)
• Verification — AI events: Anthropic — Glasswing (2026-04-07) · TechCrunch — Mythos scaling (2026-06-02) · Anthropic — When AI builds itself (2026-06-04) · TechCrunch — OpenAI Pentagon terms (2026-03-01) · OpenAI — agreement post (403 on fetch) · The Intercept (2026-09-08) · DOJ — Moreno-Gama · Fortune (2026-04-14) · CBC — shooter ban · NPR (2026-04-29) · Reuters via Yahoo — second account · NewsNation on the NYT Thiel report (2026-05-30)
• Verification — labor, markets, war: Yale Budget Lab — CPS update (2026) · Stanford — Canaries (Aug 2026) · ISW (2026-09-20) · Euromaidan Press — Sapsan (2025-12-10) · Kitco / Reuters — 44 t gold (2026-07-21) · NPR — Zelensky at UNGA (2025-09-24) · Fortune — 51st state (2025-01-08) · CBC — 51st state

migrated v1→v2 2026-09-21 · backup: _archives/diane-francis.html.bak-20260921 · converts-from: people/diane-francis.md · AI & Society domain