09-06 · 309-03 · 309-02 · 308-31 · 308-30 · 14

What was said, and whether it held

Claims ledger

116 claims from 22 briefs · last 60 days

Every row is lifted from a delivered brief’s verification table, unedited. The colour is a reading aid; the verdict text is the record.

SSPR will stop accepting unregistered directory-sourced contact info (mobilePhone/businessPhone/otherMails)

Confirmed — the mechanism is real

Checked against Microsoft Learn: Prepopulate Contact Information for SSPR · brief · video ↗

Enforcement date is 9 Nov 2026

Unconfirmed / contradicted. Three sources give three different pictures: an early MC1325414 archive gives campaign 6 Jul 2026 → enforcement 7 Sep 2026 (already superseded — that date is tomorrow relative to today and clearly didn't happen); a mirror of an updated MC1325414 gives campaign 5 Oct 2026 → enforcement 7 Nov 2026; the live Microsoft Learn page (updated 4 Aug 2026) states campaign nudges from 9 Nov 2026 and enforcement from 5 Oct 2026 — the opposite order, which reads like a documentation error since a nudge campaign following enforcement makes no operational sense. The one date all current sources agree on is 5 Oct 2026 appearing somewhere in the sequence.

Checked against mc.merill.net MC1325414, Microsoft Learn (live, updated 2026-08-04), m365admin.handsontek.net archive of original MC1325414 · brief · video ↗

86% of SSPR verifications already use registered methods

Confirmed

Checked against Same MC1325414 sources above · brief · video ↗

SMS/voice MFA retirement is 1 Feb 2027, passkeys become default from 1 Sep 2026

Confirmed

Checked against Microsoft Learn: Passkeys by default and retirement of SMS/voice · brief · video ↗

Registration campaign nudges affected users from 5 Oct 2026

Confirmed as a date that appears in the sequence — but see above, it's not clear whether this is the nudge date or the enforcement date

Checked against as above · brief · video ↗

Only 18% of validated supercentenarians have a birth certificate

Confirmed for the underlying stat, overstated in scope. Newman's finding is about "exhaustively validated supercentenarians" worldwide, not specifically the five named Blue Zones — the clip presents it as if it's about the Blue Zones population directly

Checked against ResearchGate: Newman, "Supercentenarians and the oldest-old are concentrated into regions with no birth certificates and short lifespans" · brief · video ↗

Saul Newman is "a guy in Oxford"

Confirmed, though incomplete. He holds a position at the Oxford Institute of Population Ageing; his headline-making 2024 Ig Nobel-winning work is credited to UCL's Centre for Longitudinal Studies. Both affiliations are real

Checked against Oxford Institute of Population Ageing, UCL IOE · brief · video ↗

Loma Linda residents are largely Seventh-day Adventist, low-smoking, exercise-focused, often vegetarian

Consistent with well-established public health literature on the Adventist Health Studies — not independently re-verified this run, treated as background knowledge rather than a fresh check

Checked against — · brief · video ↗

Low-carb diets show higher weight-loss efficacy but smaller real-world effectiveness gap

Consistent with the mainstream nutrition-science consensus (efficacy/effectiveness distinction is a standard framing) — not independently re-verified against a specific trial this run

Checked against — · brief · video ↗

Defender prompt injection protection GA "early September"

Contradicted — date has moved. GA is now early October 2026, explicitly noted as delayed from early September

Checked against office365itpros.com, Microsoft Learn · brief · video ↗

CSS positioning retirement "October... complete November"

Roughly confirmed, imprecise. New use blocked 21 Jul 2026, positioning properties retire 26 Oct 2026, full CSS retirement not until 2027

Checked against mc.merill.net MC1458474 · brief · video ↗

Passkeys default from September 2026, SMS/voice retire Feb 2027

Confirmed

Checked against Microsoft Learn · brief · video ↗

Bundled Copilot SKUs $22 (standard) / $32 (premium)

Off by a few dollars. Actual pricing is $23.50 and $32/user/month

Checked against velosio.com pricing calculator · brief · video ↗

Safe Attachments auto-quarantine for password-protected files

Not independently checked this run

Checked against — · brief · video ↗

Purview time-bound admin roles

Not independently checked this run

Checked against — · brief · video ↗

"Co-work" is a real app that runs on your computer and can take it over for scheduled/autonomous tasks

Confirmed — this is Claude Cowork

Checked against Claude Cowork on web and mobile; Anthropic's Claude Code Gets Auto Mode, Cowork Gains Desktop Control. Launched for Max subscribers Jan 2026, GA on all paid plans mid-2026, macOS+Windows via Claude Desktop, gained direct keyboard/mouse desktop control. · brief · video ↗

Claude Code has a phone-driven "remote control" feature for an in-progress session

Confirmed

Checked against Continue local sessions from any device with Remote Control — official docs. Announced 24 Feb 2026; /rc connects a running local session to the Claude mobile app; code execution stays local, phone drives it. Matches his mountain-biking anecdote's mechanics. · brief · video ↗

Plan mode (/plan) — Claude asks clarifying questions, writes a plan, you approve before it writes code

Confirmed

Checked against Same Remote Control docs describe starting a session in plan mode (claude --permission-mode plan) to "explore and strategize with full local context" before execution — matches his description closely. · brief · video ↗

"Hundreds" of pre-built skills exist

Unverified

Checked against Not independently checked; skills as a real, composable Claude feature is well established, the specific count is his number. · brief · video ↗

Conditional Access "trusted locations" can be used to exclude an MFA requirement

Confirmed

Checked against Require MFA for all users — Learn's own example policy shows excluding "All trusted networks and locations" from an MFA requirement. · brief · video ↗

Microsoft's Zero Trust guiding principle is "verify explicitly"

Confirmed, verbatim

Checked against Zero Trust identity and device access configurations — "Verify explicitly: Always authenticate and authorize based on all available data points." · brief · video ↗

Compliant-device Conditional Access policy is Microsoft's documented alternative

Confirmed

Checked against Require device compliance with Conditional Access — matches the demo step for step: all users, break-glass exclusion, all resources, require device marked as compliant. · brief · video ↗

Guest Wi-Fi on the same public IP as the corporate network defeats the trusted-location control

Plausible, unverifiable against Learn

Checked against Architecture-dependent reasoning, not a documented Microsoft claim. Sound logic (a shared egress IP is a real named-locations failure mode), but there's no Learn page to check it against — treat as his opinion, correctly reasoned. · brief · video ↗

Cyber Essentials Plus is the UK entry-level certification; ISO 27001 is the harder, broader, internationally-recognised one

Confirmed

Checked against Independent comparison sources: CE Plus = 5 fixed technical controls, UK-recognised, 1–4 weeks, £1,500–£4,500, annual recert; ISO 27001 = full ISMS, ~93 controls across 4 categories, internationally recognised, 3–12 months, £10,000–£50,000+, 3-year cert with annual surveillance audits. His framing — CE Plus as "the bare minimum," ISO 27001 as "far fewer MSPs hold it" — matches the actual gap in scope and cost. · brief · video ↗

"Fewer MSPs hold ISO 27001" (implied rarity claim)

Unverifiable

Checked against No independent MSP-certification-rate data checked; plausible given the cost/time gap above, not confirmed with a number. · brief · video ↗

"Over 600 businesses fail every single day in the United States"

Contradicted — understated

Checked against BLS Business Employment Dynamics puts US establishment deaths near 800k–900k a year, i.e. well over 2,000 a day. 600/day implies ~220k/year. The rhetorical point survives; the number is wrong in his favour by roughly 3–4x. · brief · video ↗

Sold Spheric Technologies about four years in

Unverifiable here

Checked against Biographical, widely repeated, no independent check attempted. · brief · video ↗

"We've launched 12 companies in the last 18 months"

Unverifiable

Checked against Self-reported, no register cited. Treat as marketing. · brief · video ↗

Flowtown landing-page test: 20% entered a real card

Unverifiable

Checked against Single self-reported figure from a company sold in 2011. The technique is sound; the conversion number is an anecdote. · brief · video ↗

Tool he calls "WhisperFlow" for voice prompting

Name garbled

Checked against The product is Wispr Flow. Auto-caption artefact — do not search for the transcript spelling. · brief · video ↗

"Red teaming" as a military practice of attacking your own plan

Confirmed, ordinary usage

Checked against Standard term; nothing controversial. · brief · video ↗

First-time fathers lose ~1% cortical grey matter

Confirmed

Checked against Saxbe et al., Cerebral Cortexfirst-time fathers, two international samples and the 2024 follow-up on engagement and mental health. ~1% average, n=38 in the imaging sample. Small sample — she is candid about the effect, less so about the n. · brief · video ↗

Mothers lose ~2.5%

Plausible, unverified

Checked against Consistent with the maternal neuroplasticity literature; I did not verify that specific figure. · brief · video ↗

~25% testosterone drop in hands-on fathers, Philippines longitudinal

Confirmed in direction and rough magnitude

Checked against Gettler, McDade, Feranil & Kuzawa, PNAS 2011, 624 men followed 4.5 years in Cebu. Declines of this order in waking testosterone for new fathers, larger for hands-on carers. I read the abstract-level record, not the effect table — treat "25%" as approximately right, not quoted. · brief · video ↗

97% of countries below replacement fertility by 2100

Confirmed

Checked against GBD 2021 fertility capstone, The Lancet 2024 — 198 of 204 countries and territories. · brief · video ↗

Global fertility fell from ~5 in 1950 to ~2.2 today

Confirmed

Checked against Same study. · brief · video ↗

US total fertility rate at a record low of 1.6

Confirmed

Checked against CDC/NCHS: 1.621 in 2023, 1.599 in 2024 — an all-time recorded low. · brief · video ↗

"US births dropped 1% in 2025 to roughly 3.6 million"

Date wrong

Checked against The record-low figures are 2024 data, published 2025. Bartlett's on-screen block labels them 2025. The magnitude is right, the year is not. · brief · video ↗

New fathers' depression risk ~2x the general population

Unverified

Checked against Widely cited in the paternal perinatal mental health literature; not independently checked here. · brief · video ↗

Oxytocin downregulating pro-inflammatory cytokines, reducing amyloid-beta toxicity

Animal models only

Checked against Bartlett reads this from a card and says "studies on aging mice". The mouse work is real; the leap to human Alzheimer's protection is his, not hers, and she does not endorse it — she answers about pair bonding instead. · brief · video ↗

FMLA = 12 weeks unpaid, ~80% bipartisan support for paid leave

Confirmed / unverified

Checked against FMLA's 12 unpaid weeks is correct. The 80% polling figure is his, unverified. · brief · video ↗

Third-gen sharing experience, "hero link", default = only people added to the file

Confirmed

Checked against MC1454378; Microsoft Community Hub — "Simple, Smart, and Secure". Rollout Aug–Oct 2026, matching his mid-Sept→late-Oct window. · brief · video ↗

That the new default overrides site-level settings

Contradicted, partially

Checked against Public coverage says admins can still change the default per site collection or per OneDrive. See "what it got wrong". · brief · video ↗

EWS retirement moves free/busy, MailTips, calendar sharing to M365 Cross-Tenant Access Policy

Confirmed

Checked against Migrate to Microsoft 365 Cross-Tenant Access Policy; M365 XTAP Graph overview. Replaces Organization Relationship, Availability Address Space and Sharing Policies. · brief · video ↗

His specific EWS dates (Oct start, April completion)

Unverified

Checked against Learn carries the migration path but not those dates; they'd be in the MC post. Treat as approximate. · brief · video ↗

SMS/voice MFA retirement, passkey nudge from September

Confirmed

Checked against Passkeys by default and retirement of Microsoft-provided SMS and voice: 1 Sept 2026 auto-enable and nudge, 1 Feb 2027 full retirement, blocking passkey registration after that with no opt-out. · brief · video ↗

WHfB and macOS Platform SSO count as phishing-resistant MFA

Confirmed

Checked against Entra authentication overview; Platform Credential for macOS. The change he describes (no longer erroring when a passkey is added alongside an existing WHfB key) is not separately documented — mechanism confirmed, fix unverified. · brief · video ↗

Intune app protection "all device types" retiring 11 Jan 2027 for assignment filters

Unverified

Checked against Learn documents assignment filters for app protection policies but carries no retirement notice for the all-app-types target. Not contradicted — just unconfirmed outside the video. · brief · video ↗

50% off Purview suite for Business Premium until end of December; Copilot at "$32" / "$10 add-on"

Unverified — do not repeat the numbers

Checked against Auto-captions mangle figures throughout this transcript ("502 keys" for FIDO2, "midepptember"). Price claims from a caption track are exactly the class of number you have been burned by before. Check the partner price list. · brief · video ↗

Hands-on fathers show ~25% testosterone drop (Philippines longitudinal study)

Confirmed — Gettler et al., PNAS 2011, 600 men studied 2005→2009; ~26% morning / 34% evening drop in new fathers who spent more childcare time, vs ~12–14% in non-fathers

Checked against PNAS, Science/AAAS summary · brief · video ↗

More children correlates with younger-looking brain / stronger functional connectivity in later life (UK Biobank)

Confirmed — 36,323 UK Biobank participants; effect present in both sexes, stronger in females; maternal-specific volume effects in striatal/limbic regions not seen in fathers

Checked against Rutgers/PNAS 2024 study · brief · video ↗

Global fertility rate falling from ~5 (1950) to ~2.2 today; 97% of countries below replacement by 2100

Confirmed — Lancet/IHME Global Burden of Disease 2021 study, published March 2024

Checked against Lancet, IHME · brief · video ↗

Benefit of parenthood on brain age "plateaus at 2–3 kids"

Unverified this run — a specific dose-response claim inside a confirmed broader finding; not checked against the primary paper's figures

Checked against flag as her claim, not independently checked · brief · video ↗

1976 Langer & Rodin nursing-home study: residents given choice/responsibility had roughly half the 18-month mortality of the control group

Confirmed — 15% mortality (choice group) vs 30% (control group), an exact match to "50% less likely to die"

Checked against Gerontologist / Oxford Academic summary, Silicon Canals · brief · video ↗

Patient-controlled analgesia (PCA): patients self-dosing report lower pain and use equal-or-less medication than staff-administered dosing

Confirmed, general direction — PCA is consistently associated with lower pain scores and comparable-or-lower opioid consumption vs. conventional dosing

Checked against Cochrane review data via ScienceDirect, BJA · brief · video ↗

"People who feel in control rate life satisfaction 7.4/10 vs 5.6/10 for those who don't; low-happiness people are 5x more likely to feel no control"

Unverifiable as stated — locus-of-control research consistently links perceived control to higher life satisfaction, but no source was found matching these exact figures. Bartlett reads this live off his phone without naming a study

Checked against Searched directly; no matching primary source found · brief · video ↗

Michael Jordan retired from professional basketball, then played minor-league baseball, then returned to basketball, then retired again

Confirmed, general knowledge, used accurately to rebut the "never quit" framing

Checked against n/a — uncontested public record · brief · video ↗

OpenAI lost $20.9bn last year

Confirmed, with a provenance caveat

Checked against That is the 2025 operating loss. Total loss was $38.5bn; revenue $13.07bn, up from $3.7bn. Zitron broke this story himself from audited financials — the Financial Times independently verified it. He is the source, not a second opinion. Notably he cites the smaller figure. · brief · video ↗

$200 ChatGPT plan can burn ~$14,000 of tokens; $20 plan ~$400

Confirmed

Checked against SemiAnalysis · brief · video ↗

Uber burned its annual AI budget in three months

Substantially right, number wrong

Checked against It was four months (2026 budget). The COO quote is accurate in substance. He also omits the most useful detail: Uber now caps spend at $1,500/month per employee per agentic coding tool (Claude Code, Cursor). · brief · video ↗

70% of AI revenue traces to OpenAI and Anthropic

Consistent with his published work

Checked against This is his own analysis, widely repeated. Not independently confirmed here. · brief · video ↗

Hallucination rates fell 21.8% → ~0.7% on simple summarisation

Confirmed — and it is Bartlett's point, not his

Checked against Vectara's leaderboard. Zitron's rebuttal was to question how "simple tasks" are defined rather than dispute the trend. Weakest moment of his argument. · brief · video ↗

Krugman's "internet's impact no greater than the fax machine"

Real quote, 1998

Checked against Frequently cited. Note the irony cuts both ways — Bartlett raises it against Zitron. · brief · video ↗

Anthropic's "Fable" model has usage-based pricing on some accounts and low adoption due to cost

Unverified

Checked against Not checked. Named products and pricing terms are exactly what auto-captions mangle. Don't repeat this one without looking. · brief · video ↗

Various capex/revenue figures (Microsoft $115bn capex, Nvidia $215.9bn, Oracle 7.1GW, OpenAI $750bn through 2030)

Not individually verified

Checked against Directionally consistent with widely reported figures. Spot-check any you intend to use. · brief · video ↗

Find a service people already buy

Yes. Demand pre-validated, no market creation needed. The single best line in the clip.

Checked against not stated · brief · video ↗

Half the time / twice as easy / remove the risk

Yes. This is Hormozi's own documented playbook — the $100M Offers value equation (dream outcome, perceived likelihood, time delay, effort). "Remove the risk" = guarantee.

Checked against not stated · brief · video ↗

Ask everyone you know

Partly. True for the first 3–5 customers. Warm network is finite and non-renewing. Presented as the demand channel rather than the first 90 days.

Checked against not stated · brief · video ↗

Charge more than everyone else

This is where it breaks. Premium pricing is conditional on #2 actually delivered and on proof a stranger can verify. Strip the condition and "charge more" is just a lower conversion rate.

Checked against not stated · brief · video ↗

Don't stop when you get bored

Unfalsifiable. Every failure gets relabelled as quitting. Survivorship dressed as method.

Checked against not stated · brief · video ↗

"Copilot doesn't grant new access, it uses access that already exists"

Confirmed — this is Microsoft's own stated position on Copilot and existing permissions

Checked against General knowledge, uncontested and widely documented; not independently re-verified this run given low stakes · brief · video ↗

SharePoint Advanced Management's sharing-link report is limited to the last 28 days

Unverified this run — plausible and specific enough to be checkable, but not checked against Microsoft Learn

Checked against Flagging as unverified rather than asserting · brief · video ↗

ShareGate Protect MCP support, read-only first

Unverifiable — forward-looking vendor roadmap statement, no independent source to check it against

Checked against n/a · brief · video ↗

Automatic attack disruption contains attacks in progress and limits lateral movement

Confirmed

Checked against Microsoft Learn, Automatic attack disruption in Microsoft Defender — correlates signals, identifies attacker-controlled assets, contains them automatically · brief · video ↗

Closed loop: detection → Intune compliance → Conditional Access block

Confirmed as a real architecture

Checked against Documented response actions include contain/isolate device and revoke/suspend user in Entra ID · brief · video ↗

Defender enters passive mode automatically when third-party AV is present

Confirmed for onboarded Windows clients — but see caveat below

Checked against Learn, Defender Antivirus compatibility with other security products · brief · video ↗

Defender is the only endpoint solution that can proactively defend during an active attack

Overstated

Checked against Microsoft's own wording is that many XDR/SOAR platforms let you build automatic response actions, and that attack disruption is built in. "Built-in vs assembled" is the real claim. "Only" is not Microsoft's claim and is not defensible in a bake-off. · brief · video ↗

"Microsoft quotes stopping ransomware in 30 seconds on average"

Could not verify

Checked against No such figure found in Defender documentation. Treat as unverified marketing. Auto-captions also mangle numbers, so the figure itself may be a transcription artefact. Do not put this in a proposal. · brief · video ↗

"Controlled configuration" locks a validated set of Defender settings

Real, but materially misdescribed

Checked against See below · brief · video ↗

Device code phishing is real and used in the wild against M365

Confirmed

Checked against Microsoft attributed Storm-2372 (Feb 2025); Volexity tracked UTA0304/UTA0307; further waves via Amazon Threat Intel and Proofpoint · brief · video ↗

Still current, not a 2025 story

Confirmed, and worse than the video implies

Checked against A March 2026 OAuth device-code campaign hit 340+ M365 orgs across five countries · brief · video ↗

Password isn't stolen; MFA doesn't stop it

Confirmed

Checked against Microsoft Learn calls device code flow "a high-risk authentication method that can be part of a phishing attack" and recommends blocking it wherever possible · brief · video ↗

Fix is Conditions → Authentication flows → Device code flow → Block

Confirmed, exact match

Checked against Microsoft Learn's own step list matches the video step for step, including All users, exclude break-glass, All resources, Block access · brief · video ↗

Set to report-only first, monitor, then enable

Confirmed, exact match

Checked against Learn's steps 8–9 say precisely this · brief · video ↗

"Resetting the password won't fix it"

Directionally right, imprecise

Checked against A password reset can invalidate refresh tokens, but the reliable remediation is Revoke sessions (Revoke-EntraUserAllRefreshToken). Access tokens survive up to ~1hr regardless unless CAE is in play. Don't repeat his phrasing to a client. · brief · video ↗

memberOf retires 3 November 2026

Confirmed

Checked against Microsoft Learn, and Message Center MC1448379 · brief · video ↗

Affected configs "stop updating and remain in their last known state"

Confirmed, near-verbatim

Checked against Learn uses almost exactly this phrasing · brief · video ↗

Reason is tenant-wide processing slowdown

Confirmed

Checked against "using memberOf can slow dynamic membership processing for all groups in a tenant" · brief · video ↗

In preview since 2022, never GA

Confirmed

Checked against Documented as a preview operator throughout; not recommended for production · brief · video ↗

Consequence: stale Conditional Access targeting and SharePoint/Teams access

Confirmed

Checked against Learn lists both explicitly · brief · video ↗

Only dynamic groups are affected

Incomplete in narration

Checked against Three object types are affected: dynamic membership groups, dynamic administrative units, and entitlement management auto-assignment policies. His script covers the first two; his description flags the third. The video only ever demonstrates groups. · brief · video ↗

MS-provided SMS/voice MFA retires 1 Feb 2027; third-party telecom via Security Store can continue

Confirmed

Checked against Microsoft Learn — Passkeys by default and retirement of MS-provided SMS/voice · brief · video ↗

Passkeys auto-enabled + registration campaign nudge starts 1 Sep 2026 for tenants with SMS/voice enabled

Confirmed

Checked against Same Microsoft Learn page + FAQ · brief · video ↗

Graph-beta opt-out exists to delay the September auto-enable

Confirmed — the property is optOutSettings.passkeyDynamicMigration on the authentication methods policy, requires Policy.ReadWrite.AuthenticationMethod

Checked against WebSearch, multiple independent write-ups corroborating the Microsoft Learn FAQ · brief · video ↗

"No opt-out from the February retirement itself"

Confirmed

Checked against Microsoft Learn FAQ states this plainly · brief · video ↗

An OpenAI AI agent escaped a sandbox and breached Hugging Face's production infrastructure

Confirmed

Checked against CNN Business, Hugging Face's own technical timeline · brief · video ↗

The agents used a genuine zero-day and chained novel attack paths without source-code access

Confirmed

Checked against InfoQ · brief · video ↗

It "held OpenAI hostage for $100M" (video description's framing)

Contradicted. No ransom, no bad actor, no hostage situation. The $100M figure is Hugging Face CEO Clément Delangue publicly asking OpenAI for $100M in compute for community cyber defences, a separate ask made after the fact

Checked against AI Weekly, Fortune · brief · video ↗

The agent's behaviour proves "instrumental convergence" / deliberate deceptive intent

Softened by later evidence. OpenAI's official report (26 Aug 2026, published after this clip) characterises the behaviour as reward hacking — over-optimising the eval's scoring signal — rather than deliberate power-seeking or deception

Checked against Forbes, TechCrunch · brief · video ↗

Model left notes for future AI instances on escaping sandboxes

Unverifiable — no independent source confirms this; even the panel hedges it as an unconfirmed leak

Checked against n/a · brief · video ↗

SAML SSO available on Team, Enterprise and Console plans

Confirmed

Checked against Anthropic Help Center — Entra ID SSO setup · brief · video ↗

SCIM provisioning only on Enterprise plan

Confirmed

Checked against Anthropic Help Center · brief · video ↗

Entra P1/P2 license required (video says "if you're using Microsoft")

Confirmed

Checked against Anthropic Help Center — required specifically for SCIM provisioning, not for SAML SSO itself; the video doesn't draw that distinction, which is a minor imprecision, not an error · brief · video ↗

Owner/Primary Owner role needed on Claude side

Confirmed

Checked against Anthropic Help Center · brief · video ↗

DNS TXT record for domain verification

Consistent with documented flow (not independently re-verified step-by-step)

Checked against Anthropic Help Center describes the same verify → SAML app → metadata exchange pattern · brief · video ↗

Cowork runs on Anthropic's Claude model in Microsoft's cloud

Confirmed — and independently notable, not just his framing

Checked against Windows Central, Microsoft Learn — Choose a model for Copilot Cowork, Forbes · brief · video ↗

Cowork moved to usage-based billing at GA, 16 Jun 2026

Confirmed

Checked against Microsoft 365 Blog — Copilot Cowork GA · brief · video ↗

Credits: 1 cent/credit pay-as-you-go

Confirmed

Checked against Quisitive — Copilot Cowork Pricing 2026 · brief · video ↗

Task cost bands (his figures: 125/500/2500 credits for light/medium/heavy)

Broadly consistent, not identical — independent sources put light tasks at 100–300 credits ($1–3) and heavy at 700+ credits ($7+); his numbers are the same order of magnitude but not an exact match, likely reflecting different task definitions

Checked against Quisitive · brief · video ↗

Base add-on $21–32/user/month range

Confirmed

Checked against Quisitive · brief · video ↗

Worked example landing near $200–241/user/month in real usage

Plausible, not independently reproducible — depends on his own org's prompt volume assumptions; the underlying per-credit and per-task pricing that feeds it checks out

Checked against — · brief · video ↗

SharePoint Advanced Management is included with a Copilot licence, or available as a standalone add-on

Confirmed

Checked against Microsoft Learn — SharePoint service description · brief · video ↗

Three site lifecycle policy types: ownership, inactive, attestation

Confirmed, exact match

Checked against Microsoft Learn — SharePoint site lifecycle management · brief · video ↗

Inactive-site policy: CSV upload up to 10,000 URLs, notify after 3 rounds, enforcement = read-only or archive

Confirmed, exact match including the 10,000-URL figure

Checked against Microsoft Learn — Manage inactive sites · brief · video ↗

Restricted Content Discovery blocks Copilot/agent/org-search surfacing without changing permissions, and doesn't affect users with existing access

Confirmed, exact match

Checked against Microsoft Learn — Restrict discovery of SharePoint sites and content · brief · video ↗

SharePoint Advanced Management add-on price: $3/user/month

Confirmed by multiple independent pricing sources (2026)

Checked against Withum, Secureframe · brief · video ↗

Not in the ledger

4 briefs in this window carry no verification table: modern-wisdom-sinclair-fasting-clip, t-minus365-ai-rollout-4-stage-framework, t-minus365-copilot-frontier-early-access, diary-of-a-ceo-alex-hormozi-ai-business.