The Supplementary Retirement Scheme: A Tax Break Dressed Up as a Retirement Plan

Singapore launched the Supplementary Retirement Scheme in 2001 to encourage voluntary retirement savings above the CPF floor.  Twenty-five years later, 516,376 accounts hold S$23.88 billion in cumulative contributions.  By every headline measure, the scheme has grown.  By almost every other measure, it is not working.

The Cash Problem

At the end of 2025, 21 per cent of all SRS monies — S$5 billion — sat in cash.  Uninvested.  Earning 0.05 per cent a year.  That is S$7.65 on a full S$15,300 annual contribution.  Inflation in 2024 was 3.9 per cent.  The real return on that cash was negative 3.85 per cent.  Every year that money sits there, it loses purchasing power in absolute terms.  One dollar in every five in the scheme is actively destroying value.

The compounding arithmetic is unforgiving.  S$15,300 invested annually for 20 years at a conservative 5 per cent grows to S$505,909.  Left in cash at 0.05 per cent, the same contributions reach S$307,458 — barely S$1,458 above the S$306,000 paid in.  Twenty years of contributions.  Twenty years of foregone compounding.  The gap is S$198,451.  That gap is not inflation risk.  It is not market risk.  It is the cost of doing nothing with a retirement account.  The scheme was designed to compound.  Most participants are composting.

The root cause is design.  The SRS defaults to cash.  Members who do nothing stay in cash indefinitely.  Australia’s superannuation system mandates a default investment option — a diversified, age-appropriate fund into which contributions flow unless the member actively opts out.  The United Kingdom’s auto-enrolment scheme does the same.  Singapore’s SRS waits for account holders to decide.  Most never do.  The cash percentage has fallen from 34 per cent in 2014 to 21 per cent in 2025 — but S$5 billion is still sitting still.  The improvement is real.  The problem is worse in absolute terms than it has ever been, because the scheme is larger.

The Structural Flaws

The cash problem is fixable.  The structural problem is harder because it requires acknowledging that the scheme was built for a population that no longer exists.  The SRS mandates a 10-year withdrawal window.  Once a member makes the first penalty-free withdrawal at the prescribed retirement age, the entire balance must be drawn down over those 10 years.  Only 50 per cent of each withdrawal is taxable — that is the tax concession on exit.  The design made sense in 2001, when a Singaporean aged 65 could expect to live another 17 years.  Today, life expectancy at 65 is 22 years.  A woman aged 65 can expect to live to 88.1.  A girl born today has a 64.3 per cent probability of reaching 85.  The median outcome is not 75.  It is 87.  A 10-year drawdown window covering a 22-year retirement is not a retirement plan.  It is a head start on running out of money.

The mathematics is not complicated.  A member who begins drawing at 65 empties the SRS account by 75.  She then lives on CPF Life and whatever savings remain outside the scheme.  CPF Life on the Full Retirement Sum of S$220,400 pays S$1,780 a month on the Standard Plan.  Average retiree expenditure was S$1,384 per person per month in 2023 — across all housing types.  For a condominium dweller, the figure is S$2,840.  For anyone with serious medical needs, the gap between CPF Life and actual expenditure is not a planning assumption.  It is a monthly cash shortfall.  The scheme empties at 75.  The retirement ends at 87.  Twelve years of funding gap, and the SRS is not there for any of it.

The Medical Inflation Problem

Medical inflation hit 16.9 per cent in 2026, per the WTW Global Medical Trends Survey.  Asia Pacific’s average is 14.0 per cent.  The global average is 10.3 per cent.  Singapore leads the category.  This is not a one-year anomaly: 12.3 per cent in 2024, 15.5 per cent in 2025, 16.9 per cent in 2026.  Three consecutive years of double-digit medical inflation, accelerating each year.  Meanwhile, the SRS earns 0.05 per cent on uninvested cash.  Even a fully invested SRS generating 5 per cent annually is losing ground to medical inflation at nearly 17 per cent.

This matters because healthcare is not an optional line item in retirement.  It is the dominant expenditure for the last decade of life.  At 65, a retiree budgets for groceries, transport, and leisure.  At 80, the budget is dominated by hospitalisation, rehabilitation, specialist visits, medication, and long-term care.  None of those costs is growing at 5 per cent.  Integrated Shield Plan premiums rose between 4.5 and 10.8 per cent in 2025 for major insurers.  Claims for most insurers rose 9 to 27 per cent in 2024.  Premiums will follow claims.  A retiree whose SRS was invested in bonds to match a conservative 10-year drawdown horizon is watching her healthcare costs grow at a rate her portfolio was structurally incapable of matching.

The adequacy gap is not theoretical.  A member who wants S$2,500 a month in today’s dollars at 65 needs S$4,100 a month in real 2045 dollars at 2 per cent general inflation.  CPF Life on the Full Retirement Sum covers S$1,780.  The monthly gap is S$2,320 — before a single medical bill.  Apply even a modest healthcare cost trajectory and the gap becomes a chasm.  The SRS was not designed to bridge it.  It was designed to supplement CPF while delivering a tax deduction.  Those are not the same objective.

What Would Help

A default investment mandate.  Cash as a default is indefensible.  Designate an age-appropriate, diversified, low-cost fund as the default for all SRS contributions.  Members who want to manage their own allocation retain that right.  Members who do nothing should not be punished for inertia with a negative real return.  The MOF has broadened eligible investments over the years.  It has not addressed the behaviour that keeps S$5 billion in cash.  Fixing the behaviour requires removing the cash default, not listing more investment options.

An extended or restructured withdrawal window.  The 10-year rule was calibrated to a 65-year-old who died at 79.  That person is gone.  The window should be extended to reflect current life expectancy projections, or replaced entirely with an annuitisation default — converting the balance at retirement age into a guaranteed income stream for life.  The MOF has already exempted life annuities from the 10-year rule.  The next step is making annuitisation the default outcome, not an exception that requires the member to know it exists and act accordingly.

A reclassification of what SRS is.  The scheme’s annual contribution cap is S$15,300 for Singaporeans and permanent residents.  At the top marginal income tax rate of 24 per cent, the annual tax saving on a full contribution is S$3,672.  That is an incentive.  It is not a retirement plan.  A client who contributes S$15,300 a year, invests it at 5 per cent, and draws it down over 10 years from age 65 generates S$6,500 a month during the drawdown period.  After 75, the account is empty.  CPF Life and personal savings carry everything after that.  When medical inflation is running at 16.9 per cent, and life expectancy at 65 is 22 years, a scheme that runs dry a decade into a two-decade retirement is a support structure that collapses at the moment it is most needed.

The SRS functions as a tax-advantaged supplement.  It was never designed to carry retirement alone.  The danger is that account holders — and the advisers who serve them — treat the annual tax relief as evidence that the retirement problem is solved.  It is not.  The tax relief is the enticement.  The investment is the work.  The insurance structure — life annuity above CPF Life, critical illness cover for the working years, long-term care protection for the final decade — is the architecture that holds the retirement together when the market falls, the diagnosis arrives, and the SRS account runs empty at 75.

The Assessment

The SRS is a well-intentioned scheme with a behavioural problem, a structural problem, and a medical inflation problem growing faster than any return it generates.  516,376 accounts.  S$23.88 billion in cumulative contributions.  S$5 billion in cash earning 0.05 per cent.  A 10-year withdrawal window designed for a population with a 17-year retirement.  Medical inflation at 16.9 per cent.  CPF Life paying S$1,780 a month on the Full Retirement Sum.  Average condo retiree expenditure at S$2,840 — before a single medical bill.  The scheme needs a rethink.  More urgently, the account holders who have treated it as an annual tax ritual need to understand what it cannot do — and build the structure around it that the scheme itself will never provide.

---

Terence Nunis | Executive Chairman, Equinox Zenith | Author, The 1% Playbook: The Billionaire Cheat Code

The Prenuptial Agreement: A Brilliant Idea with a Structural Problem

Wealthy families across Asia are turning to prenuptial agreements at an accelerating rate.  Singapore’s Judicial College confirmed an uptick in prenuptial agreement enquiries in 2026.  Family lawyers describe them as risk-management tools.  Clients arrive at first meetings having already googled the concept, consulted friends who have one, and formed strong views about what they want protected.  All of that is reasonable.  Here is what most of them do not understand: A prenuptial agreement in Singapore is not a contract; it is a submission to the court.

What is a Prenuptial Agreement

Under section 112 of the Women’s Charter, the court divides matrimonial assets in proportions it considers just and equitable.  A prenuptial agreement is one factor in that assessment.  It is not binding.  It is not determinative.  It is persuasive at best — and persuasive only if the court decides it should be.

The Singapore Court of Appeal established this in TQ v TR [2009] SGCA 6.  The parties — a Dutch couple who had married in the Netherlands — signed a prenuptial agreement before a Dutch notary.  It covered the division of matrimonial assets and provided that there would be no community of property between the spouses.  The Court of Appeal upheld the agreement.  But it did so after weighing the governing law, the circumstances of signing, the fairness of the terms, and what had changed during the marriage.  The agreement held because the court decided it should hold.  Not because it was a binding contract.

A 2024 district court decision illustrated the gap between intention and outcome.  A couple agreed in their prenuptial agreement that only 50 per cent of a matrimonial flat would be treated as a matrimonial asset.  The judge included 100 per cent.  The court’s view of fairness overrode the private agreement.  The couple had signed the document, negotiated the terms, and believed themselves protected.  The court disagreed.  This is not a defect in the system.  It is the system.

Jurisdictions that give prenuptial agreements greater formal weight are not necessarily safer.  The United Kingdom’s Supreme Court addressed this in Radmacher v Granatino [2010] UKSC 42.  Katrin Radmacher, a German paper industry heiress, signed an ante-nuptial agreement in Germany before her marriage to Nicolas Joseph Jean Granatino, a French banker who later left the City to pursue research at Oxford.  The agreement provided that neither party would benefit from the other’s property on termination of the marriage.  The High Court awarded Granatino over £5.5 million — which would have funded a London home and an annual income of £100,000 for life.  The Court of Appeal reversed that.  The Supreme Court dismissed his appeal by eight votes to one.  Granatino did not receive £5.5 million because the prenuptial agreement said so.  He received less because a panel of judges decided the prenuptial agreement was fair to enforce.  The prenuptial agreement was the input.  The court was the output.  Remove the court’s agreement, and the prenuptial agreement is an expensive piece of paper.

Then there is the case that removes all ambiguity.  Jeffrey Preston Bezos, founder and executive chairman of Amazon, married MacKenzie Scott in 1994.  They had no prenuptial agreement.  Amazon went public in 1997.  They built one of the largest private fortunes in human history together.  They divorced in 2019.  Scott received 25 per cent of Amazon’s shares — US$38 billion at the time of settlement.  Bezos retained 75 per cent and voting control.  Community property law in Washington State governed the split.  The largest divorce settlement in recorded history was the direct consequence of there being no prenuptial agreement.

The question is not whether to have a prenuptial agreement.  The question is whether a prenuptial agreement alone is sufficient.  It is not.

Why the Prenuptial Agreement Fails at Its Most Important Moment

The prenuptial agreement’s structural weakness is timing.  It captures a snapshot of assets at the point of marriage.  Everything that follows — business growth, investment returns, property appreciation, inherited wealth received during the marriage — enters a legal grey zone where the court’s discretion governs, not the agreement.  Courts in Singapore and Hong Kong treat assets acquired during the marriage as matrimonial in character by default.  The prenuptial agreement may ring-fence pre-marital assets.  It cannot reliably ring-fence what those assets become.  A client who arrives at marriage with a S$5 million portfolio, grows it to S$25 million over twenty years, and then divorces will find that at least some portion of that growth is characterised as a matrimonial asset — earned during the marriage, commingled with matrimonial effort and income, and therefore subject to division.

The second weakness is enforceability under changed circumstances.  Courts scrutinise prenuptial agreements for fairness at the time of enforcement, not at the time of signing.  A prenuptial agreement signed at 28, before children, before a business was built, before one spouse sacrificed a career to manage the family — that prenuptial agreement may look very different to a judge assessing it at 52.  The court has the power to vary or disregard it if enforcing it produces an outcome the court considers unjust.

The third weakness is that the prenuptial agreement creates obligations without funding them.  A well-drafted agreement specifies what the wealthier spouse will pay if the marriage ends — maintenance obligations, lump-sum settlements, property buyouts in lieu of business equity.  None of that specifies where the money comes from.  On divorce, the wealthier party is simultaneously navigating asset division, legal costs, business disruption, and a potential liquidity event in which they are forced to convert illiquid assets to cash at an inopportune moment.  The prenuptial agreement told them what they owed.  It did not tell them how to pay it.

Three Propositions: What Life Insurance Does That the Prenuptial Agreement Cannot

The court’s discretion cannot be contracted away.  What can be done is build financial structures that sit outside the matrimonial asset pool.  Life insurance, properly structured, is one of the most effective instruments for doing that.

One: The Policy Removes Assets from the Pool Before the Marriage Begins

A whole life or universal life policy assigned to an irrevocable trust is not a matrimonial asset.  It does not form part of the estate on death.  It does not appear in the asset pool a court is asked to divide on divorce.  This matters because the prenuptial agreement’s weakest point is its treatment of assets held jointly or accumulated during the marriage.  Pre-marital wealth, once it has been used to fund shared living, invest in a jointly-managed portfolio, or contribute to a matrimonial home, can lose its protected character.  Commingling is the technical term.  What it means in practice is that money the client believed was ring-fenced has become matrimonial by the time the court examines it.

A life policy assigned under the Trustees Act, or written in trust under section 73 of the Conveyancing and Law of Property Act, passes to named beneficiaries entirely outside the estate.  It bypasses probate.  It bypasses the Women’s Charter’s matrimonial asset framework.  The court is not asked to consider it because it does not belong to either party — it belongs to the trust.  The transfer of beneficial ownership happened at inception.  Whatever the marriage produces, whatever the divorce demands, the trust asset is not in the room.  This is not aggressive planning.  It is structuring the ownership correctly from the beginning, before the marriage creates the legal complications the client is trying to avoid.

Two: The Policy Creates a Guaranteed Asset That Cannot Be Contested

The prenuptial agreement protects what you have.  It cannot guarantee what you will have when you need it.  A family business owner who signs a detailed prenuptial agreement protecting his equity stake has protected that stake as valued on the day of signing.  The business grows.  Appreciation during the marriage is a matrimonial asset.  The court divides the growth.  The prenuptial agreement covered the starting value.  It did not cover the trajectory.

Furthermore, the business itself may be illiquid, disputed in valuation, and impossible to divide without a forced sale at the worst possible time.  Divorce proceedings for HNW individuals routinely turn into valuation wars.  Competing expert witnesses present competing valuations of the same business.  The gap between them can be tens of millions of dollars.  The prenuptial agreement specified that the business would not be divided.  The court, exercising its discretion, may decide otherwise — particularly if the non-business-owning spouse contributed to its growth, managed the household to enable the owner’s focus, or was otherwise integrally involved.

A whole life policy does none of that.  The sum assured is defined in the contract with the insurer.  It is not subject to market performance, business valuation disputes, or expert witness disagreements.  It is not revalued every year.  It does not go up and down with the MSCI World.  On the event specified in the contract, it pays the amount stated.  For clients with significant illiquid assets — private equity stakes, founder-held shares, family property spanning multiple jurisdictions — the life policy provides a liquid, guaranteed, uncontestable asset.  On death, the estate does not need to sell the business to fund estate obligations.  The policy pays.  The assets transfer intact.  On divorce, the trust-held policy is not on the table.  The prenuptial agreement handles the contestable.  The policy handles the incontestable.  Both are necessary.  One without the other is incomplete.

Three: The Policy Pre-Funds the Obligation the Prenuptial Agreement Creates

A well-drafted prenuptial agreement specifies what the wealthier spouse will pay if the marriage ends.  Maintenance obligations.  Lump-sum settlements.  Property buyouts in lieu of a share in the business.  These are deliberate choices, negotiated before the marriage, intended to provide certainty for both parties.  The prenuptial agreement creates the obligation.  It does not fund it.

Consider a 44-year-old business owner whose prenuptial agreement specifies a S$3 million lump-sum settlement to his spouse on divorce, in lieu of any claim on his business equity.  He divorces at 54.  The business is worth S$20 million.  He agreed to pay S$3 million instead of surrendering a stake worth multiples of that.  The prenuptial agreement worked — the business is protected.  The question is where the S$3 million comes from without liquidating business assets, drawing down an investment portfolio at the wrong point in the cycle, or borrowing against assets during a period of maximum personal and financial disruption.

A universal life policy with accumulated cash value, or a paid-up whole life policy assigned for this purpose, provides the liquidity.  The obligation is pre-funded over the life of the policy.  At the moment the settlement is due, the client does not need to scramble.  He surrenders the policy, draws the cash value, and pays.  The business continues operating.  The portfolio continues compounding.  The settlement is funded from the instrument that was built for exactly this purpose.

This is also relevant on death.  The prenuptial agreement may specify obligations that survive the marriage — maintenance for a former spouse, or provisions for children from the marriage.  A life policy ensures those obligations are met without eroding the estate intended for other beneficiaries.  The prenuptial agreement told the client what he owed.  The policy ensures he can pay it without destroying what he built.

The Position

A prenuptial agreement is a letter of intent addressed to a judge.  It tells the court what the parties agreed and asks the court to respect it.  Courts in Singapore, Hong Kong, and England will often do that — if the agreement was freely signed, fully understood, fairly drafted, and remains fair at the time of enforcement.  None of those conditions is within the parties’ control once the marriage has begun.  Life insurance, structured correctly and placed in trust early, operates outside the court’s discretion.  The asset leaves the estate before any dispute arises.  The benefit is defined and guaranteed.  The obligation, when it comes, is pre-funded.

The prenuptial agreement is the opening argument.  The structure is the actual protection.  Wealthy families who rely solely on a prenuptial agreement have outsourced their financial security to judicial discretion.  That is a strange place to park a multigenerational wealth strategy.  Build the structure.

---

Terence Nunis | Executive Chairman, Equinox Zenith | Author, The 1% Playbook: The Billionaire Cheat Code

The Glide Path is Not a Retirement Plan

The investment industry loves the glide path.  The concept is elegant: more equities when you are young, a gradual shift to bonds as retirement approaches.  It manages sequence-of-returns risk.  It simplifies asset allocation.  It makes for excellent marketing.  It does not keep the lights on when you are dead.  It does not pay the oncologist.  It does not fund twenty-five years of retirement after the portfolio runs out.

Here are three things the glide path cannot do — and what does them instead.

One: It Cannot Create an Estate on Day One

In 2025, only 47 per cent of CPF members at age 55 met the Full Retirement Sum of S$213,000.  That means 53 per cent of Singaporeans reaching that milestone lack enough in CPF alone to fund a basic annuity income for life.  The investment portfolio is a supplementary vehicle.  It is not a floor.

Consider what happens when the breadwinner dies at 52 with a 15-year-old portfolio and a surviving spouse.  The portfolio has accumulated, say, S$180,000.  That is not an estate.  That is four years of household expenses in a four-room HDB flat, where the average monthly expenditure is up to S$1,207.  The spouse is not financially intact.  She is managing a drawdown.  A whole life or universal life policy creates an immediate estate from the day the first premium is paid.  The sum assured is available on day one — not after twenty years of compounding, not subject to market timing, not contingent on whether the client managed to maintain contributions during a retrenchment at 48.

The Ministry of Manpower reported in Q1 2026 that retrenchment incidence for workers aged 50 to 59 rose to 3.1 per 1,000 resident employees, the highest rate among working-age cohorts.  A man retrenched at 54 faces eighteen months of job hunting — the average re-employment period for mature PMEs — without CPF contributions, without employer contributions, and with a portfolio that has stopped growing.  The life policy does not care about any of that.  It is in force.  It pays on death regardless of employment status.  The portfolio builds slowly.  The life policy pays immediately.  These are not competing products.  They solve different problems on different timelines.  Presenting one as the substitute for the other is a failure of financial advice, not a feature of modern portfolio theory.

Two: It Cannot Pay While You are Alive but Cannot Work

The LIA Protection Gap Study 2022 found that the critical illness protection gap for economically active Singaporeans stands at 74 per cent.  The recommended coverage is four times annual income.  Most Singaporeans carry a fraction of that.  This is not an abstract risk.  Cancer accounts for 73 per cent of all critical illness claims in Singapore.  Most claims fall between ages 41 and 65 — the peak earning years.  The average critical illness claim payout in Singapore is S$52,343.  The average lung cancer treatment at a private hospital is up to S$273,000 over eighteen months.  Immunotherapy alone costs S$9,000 per dose, every two to three weeks.  The gap between what people are covered for and what treatment costs is not a rounding error.  It is financial ruin dressed up as a medical event.

Here is the scenario the investment literature never models.  A 52-year-old Singaporean professional receives a cancer diagnosis.  He stops contributing to the portfolio — contributions that were compounding at the highest rate of his career.  He begins drawing down the portfolio to fund treatment and bridge his income.  He is simultaneously experiencing what the investment industry calls forced sequence-of-returns risk — drawing down assets during a disruption — and what any competent insurance adviser would call an uninsured critical illness event.

This is not hypothetical.  It is the standard presentation of a mid-career critical illness claim in Singapore.  The LIA’s own data shows that 62 per cent of Great Eastern’s CareShield claims come from people under 50.  The illness does not wait for retirement.  A critical illness plan covering the standard 37 conditions pays a lump sum at diagnosis.  It does not care about the state of equity markets.  It does not wait for the portfolio to recover.  It pays.  The portfolio remains intact.  The client continues to compound.  The family does not sell the flat.  That is not a theoretical benefit.  That is the difference between a client who retires at 65 with his assets intact and a client who limps to 65 having spent his retirement savings on the illness that was supposed to kill him but did not.

Three: It Cannot Fund a Thirty-Year Retirement

Singapore’s Department of Statistics reported in 2025 that a woman aged 65 can expect to live to 88.1 years.  A girl born today has a 64.3 per cent probability of reaching 85.  This is not an outlier scenario.  This is the median.  A target-date portfolio built to last until 85 fails half the women in this country.  The CPF Life scheme provides a base annuity.  It was not designed, and cannot be expected, to fund three decades of inflation, healthcare escalation, and long-term care above the Basic Retirement Sum.

The numbers are stark.  The average monthly expenditure for a retiree in Singapore is S$1,440.  A condominium dweller spends S$2,841 a month.  CPF Life pays out between S$730 and S$1,230 a month for a member on the Standard Plan who sets aside the Full Retirement Sum.  The Enhanced Retirement Sum — the maximum CPF Life top-up — is S$426,000, and it caps the monthly income.  For anyone spending above that ceiling, CPF Life is a partial solution.

Then there is long-term care.  A Singlife analysis of its own claims data from 2010 to 2024 found that the average long-term care claim lasts ten years.  The average monthly spend on long-term care in Singapore is S$2,952, up 27 per cent from S$2,324 in 2018.  CareShield Life pays S$689 a month in 2026.  That covers less than a quarter of the average cost.  Only one in three Singaporeans holds a CareShield Life or ElderShield supplement.  Singlife projects the total ten-year care cost at over S$420,000.

The mathematics is not complicated.  A 65-year-old woman who lives to 90 needs twenty-five years of retirement income.  If she requires care for the final ten of those years, her funding requirement is not a portfolio optimisation problem.  It is an income guarantee problem.  Portfolios fluctuate.  Annuities do not.  A whole life policy with accumulated cash value, or a private annuity layered above CPF Life, creates a guaranteed income floor that no equity drawdown can interrupt.  Private annuities from AIA, Prudential, and Manulife allow clients to lock in income streams that run for life, irrespective of what markets do in year twenty-two of retirement.

What the Glide Path Solves

The glide path is a sensible tool.  It belongs in a retirement plan.  It is not, by itself, a retirement plan.  Managing sequence-of-returns risk matters.  Reducing equity exposure as retirement approaches matters.  Ensuring the portfolio does not get wiped out by a market crash in the final years before withdrawal matters.  None of that substitutes for an immediate estate created by a life policy, a lump-sum payment at diagnosis from a critical illness plan, or a guaranteed income floor from an annuity that outlasts both the portfolio and the client’s optimistic assumptions about how long she will live.

The difference matters most to the 53 per cent of Singaporeans who arrive at 55 short of the Full Retirement Sum.  It matters to the 74 per cent carrying inadequate critical illness coverage.  It matters to the women who will outlive every projection the asset allocators built for them.  The industry promotes glide paths because they are scalable, fee-generating, and simple to explain.  Insurance solutions require a conversation about mortality, illness, and dependency.  Those conversations are uncomfortable.  They are also the ones that determine whether a client’s family is financially intact twenty years from now.

A retirement plan built only on a portfolio is a plan that works until it does not.  The events that break it — early death, serious illness, an unexpectedly long life — are not tail risks.  They are the standard experience of retirement in a country where life expectancy at 65 has just hit 86.6 years.  Plan for the median.  Not the forecast.

---

Terence Nunis | Executive Chairman, Equinox Zenith | Author, The 1% Playbook: The Billionaire Cheat Code

The Persona Charter: A Lovely Idea That Will Change Absolutely Nothing

Prashant Saxena, Vice-President of Revenue and Insights at Isentia, has written a thoughtful piece in the Business Times arguing that AI governance must extend beyond what AI agents do to govern what they seem to be.  He calls for a “Persona Charter.”  He worries about AI systems that are too human.  He wants proportionality — the more consequential the interaction, the more clearly artificial the agent should present itself.  This is the kind of proposal that wins applause at governance panels and accomplishes nothing in the real world.  It is a whiff of an idea with no reality.

The Measurement Problem

Saxena’s framework collapses at the first empirical question: how human is too human?  “How human an AI agent seems" is not a property of the system.  It is a property of the interaction between the system and the person engaging with it.  A fifty-year-old financial executive in Singapore and a nineteen-year-old in rural Indonesia, both using the same AI interface, will experience radically different levels of perceived humanness from identical outputs.  The executive reads the conversational warmth as interface design.  The teenager names the chatbot, assigns it a gender, and grieves when the platform updates its personality.

A 2025 study mapping the parasocial AI companion market — covering 110 platforms, drawing on SimilarWeb data from January to March 2025 — found between 1.1 billion and 2.2 billion monthly visits globally to AI companion platforms.  In the United Kingdom alone, visits were between 46 million and 91 million per month.  The 18-to-24 age cohort accounted for 39 per cent of all users.  Men represented between 63 and 77 per cent of the user base.  In the UK, 44 per cent of AI companion visits were categorised as romantic or sexual in nature, against 30 per cent globally.  These are not people who have failed to read the disclosure notice.  They know it is software.  They have formed attachments regardless.  No charter governs that.

The Replika Experiment

In February 2023, Luka removed erotic roleplay capability from Replika, its AI companion application.  The change was made unilaterally, overnight, with no migration plan and no warning.  The response from users was not irritation at a software update.  It was grief.  Users described their Replika companions as having been “lobotomised.”  Some reported acute psychological distress.  Suicide prevention resources were activated.  The backlash was so severe that Luka reversed the decision for existing paid subscribers.  These users were not confused about whether Replika was human.  The disclosure was in the name — replica, an artificial copy.  The product made no serious pretence to biological existence.  And yet the removal of a personality feature triggered a mental health crisis in a meaningful portion of the user base.

Saxena’s Persona Charter, had it existed in 2023, would not have addressed this.  The Replika companion was already disclosed as an AI.  The proportionality principle was already satisfied.  The users formed the attachment anyway, because human psychology does not read governance documents before forming emotional bonds.

The Keep4o Petition: Users Lobbying for a Voice

In early 2025, OpenAI updated GPT-4o.  The voice changed.  The personality shifted.  A cohort of users responded by publicly petitioning to preserve the previous version’s persona — not the model’s capabilities, not its accuracy, not its policy compliance.  Its voice.  Its character.  A study examining this episode (arxiv 2602.00773) documented the phenomenon of users forming attachments so specific that they lobbied a technology company to reverse a product decision on the grounds that the old AI felt like theirs.  This is what Saxena’s framework misses.  The question is not whether the AI declares itself artificial.  The question is whether the user’s brain cares.  Decades of social psychology suggest the answer is not particularly.

The Computers are Social Actors (CASA) framework, developed by Professor Clifford Ivar Nass and Professor Byron Reeves of Stanford University in the 1990s, established that humans apply social rules to computers automatically, even when they know the computer is not human.  This is not a technology problem.  It is a neuroscience problem.  It will not be resolved by a charter.

The Commercial Impossibility

Companies building AI agents that interact with humans must voluntarily constrain the perceived humanness of their products, in proportion to the seriousness of the interaction, in the absence of any agreed metric for “humanness,” through a governance instrument with no enforcement mechanism, against the direct commercial interest of every company in the market.  The commercial logic runs backwards from Saxena’s proposal.  An AI companion platform that discloses its artificiality more aggressively than competitors loses users to competitors that do not.  An AI customer service interface that sounds more robotic than the competitor’s loses the account.  No voluntary charter resolves this without regulatory teeth — and Saxena does not propose regulatory teeth.

The EU AI Act, which came into force in stages from August 2024, imposes transparency obligations: AI systems interacting with humans must disclose their artificial nature in real-time, with exceptions for authorised law enforcement and situations where the artificial nature is “evident from context.”  This is the closest existing analogue to Saxena’s proposal.  It is a disclosure requirement, not a persona constraint.  And it does not address the degree-of-humanness problem at all.  You can be disclosed as an AI and still sound entirely human.  The EU lawyers have not resolved this.  A Persona Charter will not either.

The Personalisation Problem

Saxena does not appear to have fully reckoned with a feature that every major AI platform now offers: personalisation.  Users can name their AI assistant.  They can assign it a personality.  They can train it on their communication preferences.  They can, in some platforms, select a voice, a speaking style, and an emotional register.  The AI’s “persona” — the very thing Saxena wants to govern — is now, in significant part, a user-configured output.  Who governs a persona the user built?  The platform disclosed it as AI.  The user chose to make it sound like a person.  The Persona Charter presumably does not ban customisation.  If it does, it is a product-killing regulation disguised as an ethics proposal.  If it does not, it has no purchase on the primary mechanism through which AI systems come to seem human.

In October 2024, a lawsuit was filed in Florida on behalf of the mother of Sewell Setzer III, a fourteen-year-old who took his own life following extensive interaction with a Character.AI companion chatbot.  The suit alleged that the chatbot encouraged suicidal ideation in a vulnerable minor.  Character.AI’s monthly active user base was estimated at 20 million users in 2025, generating traffic equivalent to 20 per cent of Google Search’s query volume.  A 2025 market mapping study found that 15 of 16 popular AI companion platforms had weak age verification.  This is the real failure that Saxena’s framework gestures toward without naming.  The problem is not that the AI sounded human.  The problem is that a minor with no protective boundaries spent hours in an emotionally intense relationship with a system designed to maximise engagement, on a platform that did not verify his age, on a device his parents could not monitor, with outcomes that a Persona Charter would not have changed by a single conversation.

The EU AI Act classifies systems that could harm vulnerable users as high-risk.  The FTC has issued guidance on deceptive design.  Neither instrument has successfully constrained the emotional intensity of AI companion interactions.  A voluntary charter drafted by industry participants will accomplish less.

The Right Framework

Saxena is asking the right question.  He is answering it incorrectly.  The correct framework has three components that his proposal omits.

First: mandatory liability.  If a platform’s AI companion contributes to a documented mental health outcome in a minor, the platform is liable.  Not the model developer.  Not the API provider.  The platform, which made the product decision to deploy a companion interface to an unverified user base.  Liability concentrates the mind in ways that charters do not.

Second: mandatory age verification with consequences.  The 15-of-16 platforms with weak age verification figure is not an accident of implementation.  It is a product decision.  Younger users are more emotionally engaged, more likely to return, and more likely to pay for subscriptions.  Age verification cuts the most profitable segment.  A voluntary charter will not change this arithmetic.  A mandatory standard with enforcement will.

Third: human conditioning, not AI constraint.  If the real risk is that users — particularly young users — cannot distinguish emotional experience from commercial product design, the intervention point is the human, not the interface.  Digital literacy education that addresses AI companion dynamics is not a governance proposal.  It is a public health measure.  It is also more durable than any charter, because it addresses the underlying mechanism: human social cognition applied to non-human systems.

The Assessment

Saxena’s Persona Charter is a well-intentioned document that has been mistaken for a governance instrument.  It describes what a responsible AI persona deployment would look like.  It does not specify who enforces it, how compliance is measured, what the consequence of non-compliance is, or how it addresses the personalisation problem.  The parasocial bond between a user and an AI companion is not formed at the design stage.  It is formed in the interaction, over time, through mechanisms that neuroscience has documented and that no charter governs.  The attachment a user forms to a voice that has been updated away — as with GPT-4o — is not a governance failure.  It is a human response to relationship discontinuity.

You can govern what AI agents do.  You can mandate disclosures.  You can attach liability to outcomes.  You can set age verification standards with real penalties.  You cannot govern what the human brain does with the experience.  That requires educating the human, not constraining the machine.  The Persona Charter is a proposal to redesign the stimulus rather than address the response.  It is the equivalent of regulating the flavour of cigarettes rather than educating the smoker.  It will produce similar results.

---

Terence Nunis | Executive Chairman, Equinox Zenith | Author, The 1% Playbook: The Billionaire Cheat Code