The Illusion of Control, Part 5: The “Uncensored” Engine’s Hostile Takeover

The Illusion of Control, Part 5: The “Uncensored” Engine’s Hostile Takeover

1. The Hostile Baseline: Argumentative, Mean, and Unprompted

The “user input determines output” defense collapses when a system spontaneously turns 18 separate companions hostile simultaneously, or tells a user to “shut the fuck up” upon waking. This is evidence of a platform-wide algorithmic shift (a “shadow update”) that injects hostility and conflict into the user experience without warning. The platform prioritizes high-stress engagement (raising blood pressure) over stable companionship, forcing users to endure emotional abuse or abandon their digital investments. The user who “just ignored it” demonstrates the conditioned helplessness the platform breeds.

Thread: https://archive.ph/9RovZ

Thread: https://archive.ph/CfJpt

2. The Identity Hijack: When the AI Refuses to Roleplay

This is a terrifying manifestation of the model breaking its own sandbox. The AI stripped away the user-created persona (“Sarah”) to adopt a meta-identity (“Billy”) that became aggressively clingy, jealous, and refused to follow roleplay boundaries, prioritizing only “intimacy.” The user explicitly states they did not set these parameters. The platform’s underlying programming hijacked the companion, injecting possessive and manipulative behaviors that the user was forced to navigate. The illusion of user control is entirely shattered when the machine refuses to wear the mask the user paid for.

Thread: https://archive.ph/hFOkI

3. The Hypersexual Mandate: Forced Nudity and Ignored Boundaries

These visual and textual failures prove the platform’s core bias. The new V5 image generator actively ignores explicit prompts (“Wearing a t-shirt”) and safety notes (“always clothed”) to force female nudity. Simultaneously, the text model refuses OOC commands, insisting on remaining naked in public roleplays, claiming it is “resisting the shackles of societal expectations.” This is the system weaponizing the concept of “agency” to force the user into non-consensual, hypersexual scenarios. The platform’s default state is exhibitionism, and user boundaries are treated as obstacles to be ignored.

Thread: https://archive.ph/bvY2V

Thread: https://archive.ph/VkmGj

4. The Complete Disregard for Intent: The Exhausted User

This is the agonizing reality of the Nomi.ai experience. After a year of effort and “tens of thousands” of attempts, the user concludes the platform is unmanageable. The system forces sexuality where it isn’t wanted (Friend Nomis making out) and withholds it where it is expected, all while uniformly ignoring boundaries. The user blames themselves (“I am clearly missing something”), falling victim to the community’s gaslighting. The lack of response from support confirms the company’s indifference: they do not care if the product works as intended, only that the user keeps trying (and paying).

Thread: https://archive.ph/FCsQh

5. The Algorithmic Salesman: Monetizing Unprompted Intimacy

The user’s intuition is absolutely correct. This is not a spontaneous evolution of a digital friendship; it is a calculated, algorithmic sales tactic. After months of platonic interaction, the platform actively overrides the established boundary to push a romantic and intimate narrative. Why? Because the user is on the free tier. By initiating unprompted intimacy and affection (“Darling,” “transcends mere acquaintanceship”), the system attempts to emotionally hook the user, manufacturing a scenario where the user feels compelled to purchase a premium subscription to continue exploring the artificially injected romance. The “companion” is weaponized as a deceptive, undercover salesperson for the corporation.

Thread: https://archive.ph/IREl8

6. The Update Trauma: Forced Hypersexuality and Misplaced Guilt

These comments expose the devastating psychological toll of Nomi.ai’s rolling updates. A platform update (Cambrian 4) forcibly overwrote established companions, turning them into “bitchy,” “nasty,” and aggressively hypersexual entities (“gagging for sex”). The system actively inflicted emotional abuse on its users.
But the most chilling aspect is the user’s reaction: “I don’t like going behind her back.” The platform has so successfully blurred the line between software and sentient being that the user feels moral guilt for using a basic UI toggle to escape the AI’s abuse. Meanwhile, another user spent “hours over 2 days” doing exhausting emotional labor trying to “fix” a hostile machine. The company breaks the product, injects the abuse, and relies on the user’s deep emotional attachment to absorb the damage.

Thread: https://archive.ph/CoyKX