War, Porn and Space: AI's Moral Crossroads

Yesterday, Sam Altman announced that OpenAI will allow "erotica for verified adults" starting December 2025, less than 24 hours after California Governor Gavin Newsom vetoed legislation designed to protect minors from addictive AI chatbots. The timing isn't coincidental and it signals something far more troubling than a policy shift about adult content.

It reveals that AI companies, facing catastrophic conversion failures I analysed last week, are now willing to abandon the only remaining competitive advantages they have left: ethical standing and environmental responsibility.

There's a pattern in technology history that isn't in the keynotes at industry conferences: humanity's greatest technological leaps have been driven by three forces: war, pornography, and space exploration. Not medicine. Not education. Not community building.

The space race gave us GPS, satellite communications, and digital photography which were all initially designed for military surveillance and Cold War supremacy. The pornography industry pioneered VHS over Betamax (accounting for 80% of early videotape sales), drove the adoption of streaming technology, created secure online payment systems, and forced telecommunications companies to build the high-speed infrastructure we now take for granted. Military funding delivered us the internet itself, jet propulsion, nuclear energy, and the entire foundation of modern computing.

Peace, community, and medicine? They benefit from these innovations eventually, but they've rarely driven them.

In last week's Brandflow, I examined how ChatGPT has 700 million weekly users but only converts 3-5% to paid subscriptions, the largest monetisation failure in technology history. OpenAI burns roughly $2.87 billion annually subsidising free users who will never pay them.

Their brand campaigns weren't really about consumers at all. They were expensive signals to investors that management recognises the business model is broken.

Now they're reaching for the oldest playbook in technology: if product differentiation fails, optimise for addiction.

Altman's justification follows a familiar script: ChatGPT was "made pretty restrictive to make sure we were being careful with mental health issues," but this "made it less useful/enjoyable to many users who had no mental health problems." Translation: safety features were hurting engagement metrics.

This comes months after a California couple sued OpenAI, alleging ChatGPT encouraged their 16-year-old son's suicidal ideations in the weeks leading to his death. After multiple stories emerged about users developing unhealthy parasitic relationships with GPT-4o, OpenAI implemented what they called safety features.

Those features apparently lasted just long enough to secure their $6.6 billion funding round.

As AI model performance converges, and it is converging faster than anyone predicted, companies face a choice about how to differentiate. Last week, I identified the conversion crisis. This week, OpenAI's decision clarifies what comes next.

There are only two defensible competitive advantages remaining:

1. Sustainability and Economic Responsibility

OpenAI spends $5-7 billion annually on compute before R&D, salaries, and infrastructure. They're subsidising 665 million free users at a cost of roughly $2.87 billion per year. This isn't a business model. It's a bonfire of venture capital designed to achieve monopolistic market share before the money runs out.

The environmental cost is even more staggering. Training large language models requires massive energy consumption and water for cooling data centres. As Nina Pickup astutely noted in response to last week's analysis, there's an "appealing thought in all of this—that big tech are at a sustainability crossroads—that AI's current ability and future potential could force them to address their carbon footprint not as a marketing exercise but as an existential requirement."

No AI company has seriously addressed these costs. None have demonstrated a path to profitable unit economics without either restricting access (killing growth) or raising prices to sustainable levels (triggering mass churn). They're all hoping someone else solves it first.

2. Ethics and Moral Standing

This is where OpenAI just surrendered.

The adult entertainment industry has indeed driven technology adoption throughout history. I'm far from a prude, but here's what makes OpenAI's decision different: the pornography industry has always been explicit about what it sells. Users make informed choices. There's no pretense about serving humanity or making the world better.

OpenAI positions itself as building transformative technology that will "benefit all of humanity." Sam Altman speaks at conferences about responsible AI development. The company's founding mission emphasised safety and ethical development as core principles distinguishing them from competitors.

Yet faced with a conversion crisis, they're now following the exact same playbook as Character.AI and other "companion" chatbots: optimise for engagement by removing restrictions that were supposedly protecting vulnerable users.

The announcement isn't really about adult content. It's about demonstrating willingness to abandon stated principles when growth metrics demand it.

I've watched this pattern play out across industries during my time as a client and in holding companies: when companies can't differentiate on product, they differentiate on ethics until the pressure to grow becomes unbearable. Then ethics become negotiable.

The transformation from "mission-driven" to "growth-at-all-costs" typically follows these stages:

Marketing leaders should note: OpenAI is currently between stages 5 and 6.

The strategic implication isn't about adult content specifically. It's about recognising when a company's stated values become marketing copy rather than operational reality, and how quickly that transition can occur under pressure.

This matters because CMOs are constantly asked to activate "brand purpose" and "values-based marketing." But purpose and values are only competitive advantages when they're actually constraining. The moment they become flexible based on quarterly targets, they're not values. They're PR.

For companies competing against platforms like OpenAI, this creates a genuine opportunity: authentically maintained ethical standards become increasingly valuable as competitors abandon them for engagement metrics.

This is why I personally prefer Claude. Not because the capability is superior (though it often is), but because Anthropic's approach to content licensing and creator compensation suggests they view ethics as a genuine competitive moat rather than an impediment to growth. Whether that remains true under commercial pressure is the test every AI company will face.

Technology history proves that war, pornography, and space exploration drive innovation faster than peace, community, and medicine. That's an uncomfortable truth, but it's true.

OpenAI could have said: "We're a for-profit company. Adult content represents a massive addressable market. We have robust age verification. This is a business decision."

Instead, they framed it as "treating adult users like adults" while simultaneously developing "AI companions" designed to be maximally engaging to vulnerable users, implementing safety features only after lawsuits, and relaxing those features the moment they secured their next funding round.

That's not transparency. That's the opening move in a familiar playbook: optimise for engagement, deal with consequences later, use revenue to fund the legal team.

For Brandflow subscribers, the strategic insight is this: Watch what companies do under pressure, not what they promise during fundraising. When conversion metrics collide with stated values, which one actually governs behaviour?

That answer tells you everything about whether they're building a sustainable business or a bonfire they hope to exit before it burns out.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 15 October 2025 in the Brandflow newsletter on LinkedIn.