Automated Gaslighting

We are repeatedly told that artificial intelligence is the engine of human progress and is a revolutionary tool designed to cure diseases, solve complex logistics, and elevate our daily lives. The reality, at least on the ground floor of consumer tech, is far more mundane and deeply cynical. Companies are weaponizing these massive, resource hungry models not to help us, but to build unaccountable walls between themselves and their users.

I recently spent days trapped in a Kafkaesque loop with Anthropic’s “Fin AI Agent” over a $20 charge for a Claude Pro subscription. The details of the dispute are incredibly basic: a payment failed due to insufficient funds, but the automated billing system flagged it as a “previous refund,” blocking any attempt to clear the account.

What followed was a masterclass in automated gaslighting.

I presented the AI support agent with the facts. I explained the fundamental difference between a declined charge and a processed refund. The bot, designed to mimic human reasoning, actually agreed with me. It replied, “You’re correct that a failed payment is not a refund.” Yet, in the very next breath, it admitted it lacked the security access to view my actual billing logs to verify this fact. When I requested to escalate the ticket to a human who possessed the necessary administrative tools, the digital brick wall slammed shut.

“I want to be direct with you,” the bot responded. “For individual plan accounts, AI support through this channel is the primary support structure available.”

Because I held an individual plan rather than a lucrative enterprise contract, I was deliberately denied access to human support. The system was smart enough to understand it was wrong, but explicitly programmed to be powerless to fix it—and structurally designed to prevent me from reaching anyone who could. I was eventually forced to initiate a chargeback through my bank, which resulted in Anthropic canceling my account.

This isn’t a glitch; it is the intended architecture of modern tech support.

We need to talk about the sheer absurdity of this arrangement. Anthropic is running Large Language Models that require massive data centers, immense computational power, and a staggering environmental footprint. We are burning fossil fuels, straining power grids, and contributing to a climate crisis that will eventually cost lives and displace communities.

And for what? So a tech company can use a world altering supercomputer as a digital bouncer to protect a $20 margin.

They have built a customer service system that operates purely on attrition. It is designed to exhaust the user until they simply give up and walk away. The burden of resolution is shifted entirely onto the consumer, while the company shields itself behind an algorithm that politely apologizes for its own engineered uselessness.

If this is the frontline of the AI revolution, it is not about user empowerment. It is about corporate evasion. And the collateral damage, from our collective patience to the planet itself, is a price we are all being forced to pay.

(Note: This is Part One of a three part series on the architecture of automated avoidance and the true cost of consumer facing AI.)