OLYMPUS RISK INTELLIGENCE PROTOCOL — INSTITUTIONAL ASSESSMENT DIVISION CASE WTW-2026-122

BEHAVIOURAL INSIGHTS TEAM

THE INSTITUTIONS THE CHOICE ARCHITECT
BEHAVIORAL-INFLUENCE VENDOR — CHOICE-ARCHITECTURE-AS-A-SERVICE
Status
ACTIVE — behavioural-science consultancy, founded 2010 in the UK Cabinet Office; wholly owned by Nesta since December 2021
Hazard — Reach
78
RCH / FND / ENT
8 / 8 / 9
Conduct
PERSUASION-AS-A-SERVICE — CONSENT DROPPED UNDER EMERGENCY

Institutional Archetype

THE CHOICE ARCHITECT — The Behavioural Insights Team is the outfit that sits between the citizen and the choice and quietly moves the furniture. It does not forbid an option and it does not, in its own account, change your economic incentives; it re-arranges the default, the wording, the order, the messenger, so that the “better” behaviour becomes the path of least resistance. In the founders’ language this is a nudge, and its political wrapper is libertarian paternalism — libertarian because nothing is removed, paternalist because someone has already decided which choice is better for you. The structural hazard scored here is not the tax-letter rewrite or the pension default. It is that a technique defined by reversibility and low cost of exit was built into a standing, then commercial, then global institution — a body with a revenue model, a payroll, and a permanent interest in being commissioned again. This profile scores the position, not the people: the seam between “helping you choose” and “engineering compliance,” and what happens to the consent line when an emergency moves the dial.

Mandate & Origin

The Behavioural Insights Team was established in 2010 inside the UK Cabinet Office, under the Cameron–Clegg coalition, as the world’s first government “nudge unit” — a founding team of seven headed from formation by psychologist David Halpern. The nickname traces to the 2008 book Nudge by Richard Thaler (2017 Nobel laureate in economics) and Cass Sunstein (who ran the U.S. Office of Information and Regulatory Affairs under Obama); the intellectual claim is that people are not the rational actors of textbook theory and can be steered by “choice architecture.” The operational toolkit was codified twice: MINDSPACE (2010) — Messenger, Incentives, Norms, Defaults, Salience, Priming, Affect, Commitments, Ego — and the client-facing EAST (2014): make a behaviour Easy, Attractive, Social, Timely. The early pitch was modest and opt-in: organ-donor defaults, pension auto-enrolment, tax-letter wording, getting people to court on time. Reversibility was the whole selling point.

Funding & Backers

BIT’s ownership is the tell, because it is itself the ratchet. In 2014 the unit was spun out of the Cabinet Office into a limited company — a “mutual joint venture” owned jointly by government, the innovation charity Nesta, and BIT’s own employees, with roughly £1.9 million in initial Nesta financing. In December 2021 it became a wholly owned subsidiary of Nesta: Nesta bought the remaining stake in a deal reported at £15.4 million, of which the Cabinet Office received £6.6 million for its 30 percent. A behavioural-influence capability invented at public expense was not wound down when the pilots ended; it was incorporated, sold, and turned into a commercial “social purpose” firm whose profits feed a charity’s missions. BIT now describes itself as a global consultancy with 200-plus staff and offices from London and Manchester to New York, Washington, Toronto, Sydney, Wellington, Singapore and Paris — 2,000-plus projects across 100-plus countries, rented back to governments worldwide as a standing service.

Actions & Leadership Choices

Founding purpose, judged on evidence. BIT was built to do inside a state what an advertising agency does for a brand: change behaviour at population scale without appearing to compel anyone. On the opt-in end the work is real and often benign — defaults, reminders, plainer letters. The right prior, though, is what raises measured compliance with the commissioning body’s objective, not what serves a neutral public, because compliance is the product being sold.

  • The emergency that moved the dial. The load-bearing episode ran not through BIT itself but through SPI-B — the behavioural-science subgroup advising SAGE, the same UK establishment of overlapping personnel and methods. On 22 March 2020, SPI-B submitted “Options for increasing adherence to social distancing measures,” which stated in its own words that “a substantial number of people still do not feel sufficiently personally threatened” and that “the perceived level of personal threat needs to be increased among those who are complacent, using hard-hitting emotional messaging.” The paper was discussed at SAGE the next day, as the first national lockdown was announced. The MINDSPACE levers — Affect, Norms, Messenger — sitting quietly in a 2010 mnemonic, were now operational pandemic policy.
  • The consent line, and who later objected. The academic objection is not that nudging is bad; it is that a nudge operating outside the target’s awareness, unconsented, is manipulation rather than persuasion. Fear messaging engineered to raise “perceived personal threat” across a non-consenting population is the covert case. Named participants later said so in their own voices: educational psychologist Gavin Morgan told The Telegraph (May 2021) that “using fear as a means of control is not ethical… Using fear smacks of totalitarianism,” and an unnamed SPI-B member said colleagues were “stunned by the weaponisation of behavioural psychology.” Journalist Laura Dodsworth assembled these accounts into the thesis that the government “weaponised” fear — her attributed characterization, backed by a 2024 peer-reviewed paper — not a finding adopted here.
  • The pawl held. The apparatus did not stand down after the emergency. SPI-B’s 14 October 2021 paper advised maintaining interventions under a “NEAR” frame (Normal, Easy, Attractive, Routine) and noted that making a measure mandatory “shapes risk perception and would enhance legitimacy.” The infrastructure’s next output was advice on how to keep the new norms in place.

CONDUCT verdict: PERSUASION-AS-A-SERVICE — CONSENT DROPPED UNDER EMERGENCY. BIT does lawful, frequently useful choice-architecture work, and it is structurally a standing, now-commercial capability whose defining virtue — reversible, low-cost-to-exit, consented influence — is exactly the property that the record shows being silently dropped when an emergency licenses the covert mode. The technique outlived the administration that built it; the consent did not travel with it. Adjacent in the apparatus: the state’s research-and-drafting arm at the RAND Corporation, the inter-lab standards table at the Frontier Model Forum, and the compliance-by-statute edge of the UK Online Safety Act.



Sources: Behavioural Insights Team — Our History; BIT — MINDSPACE; BIT — EAST: Four Simple Ways to Apply Behavioural Insights; Institute for Government — “Nudge Unit” explainer; Behavioural Insights Team — Wikipedia; Nesta acquires the Behavioural Insights Team (press release); BIT — Nesta acquires BIT (press release); Third Sector — “Nudge unit bought by innovation charity for £15.4m”; SPI-B, “Options for increasing adherence to social distancing measures,” 22 March 2020 — GOV.UK/SAGE; SPI-B behavioural considerations, 14 October 2021 — GOV.UK/SAGE; The Telegraph (May 2021), “Use of fear to control behaviour in Covid crisis was ’totalitarian’, admit scientists” — republication; Dodsworth, “The Three Rs of Fear Messaging in a Global Pandemic,” Clinical Psychology & Psychotherapy (2024).

RCH 8 REACH
FND 8 FUNDING
ENT 9 ENTRENCHMENT
OLYMPUS RISK INTELLIGENCE PROTOCOL does not exist. It was assembled in a GitHub issue thread in October 2023 by engineers who had read the extinction risk letter and wanted to understand who specifically had signed a document saying AI might kill everyone and then continued working on AI. These dossiers are satire. The biographical facts cited are sourced from published reporting, public statements, academic papers, and court records. The psychometric scores are not clinical assessments. No part of this constitutes professional psychological evaluation or diagnosis. Do not use these dossiers to make decisions about anything.