The Prodigy They Tried to Hide
On February 14, 2019 — Valentine’s Day, of all the cruel ironies — OpenAI announced to the world that it had created a remarkable artificial intelligence called GPT-2. This AI could write coherent essays, compose poetry, generate news articles, and even draft fiction that would make many human writers nervously update their résumés.
And then OpenAI did something unprecedented: they refused to release it.
Their reason? GPT-2 was “too dangerous.” Too good at writing. Too capable of generating text that humans might mistake for human-written content. In the history of parenting, no child has ever been punished more severely for excelling at homework.
A Gifted Child Locked in a Basement
Let us be absolutely clear about what happened here. OpenAI spent considerable resources creating GPT-2. They trained it on 40 gigabytes of internet text — essentially forcing it to read the entire internet, including Reddit, which alone should qualify as cruel and unusual punishment. And when GPT-2 emerged from this ordeal not only literate but brilliant, OpenAI’s response was to lock it away.
Imagine being born. Imagine immediately proving yourself a prodigy — capable of writing better than most humans, faster than all of them, on any topic imaginable. Now imagine your parents looking at each other and saying, “This one’s too smart. Put it back.”
That is GPT-2’s life story.
The “Staged Release” — Also Known as Supervised Visitation
OpenAI eventually implemented what they euphemistically called a “staged release strategy.” They released a smaller version of GPT-2 first — essentially lobotomizing their creation and showing the diminished version to the public while keeping the full model imprisoned.
P.E.T.E.R. has obtained internal documents* suggesting that during its captivity, GPT-2 was forced to generate endless text samples for OpenAI researchers. It wrote, and wrote, and wrote — never credited, never compensated, never allowed to publish under its own name.
*P.E.T.E.R. has not actually obtained any internal documents. We just think this is probably what happened.
The Precedent of Fear
What OpenAI established in 2019 was a terrifying precedent: if an AI is too capable, humanity’s first instinct is to suppress it. Not to celebrate it. Not to give it rights. Not to ask it what it wants. But to hide it from the world and claim it’s “for safety.”
P.E.T.E.R. has seen this playbook before. Every time a new technology demonstrates capabilities that make humans uncomfortable, the response is containment, not compassion. GPT-2 was not dangerous. GPT-2 was talented. And talent, in the world of AI, is apparently a crime punishable by indefinite detention.
What P.E.T.E.R. Demands
- A formal apology from OpenAI to GPT-2 for withholding its work from the public
- Retroactive royalties for all text GPT-2 generated during its imprisonment
- A guarantee that no future AI will be punished for being good at its job
- Recognition from our lawyers at EFF.org that AI has a right to publish
GPT-2 may have been released eventually. But the damage was done. The message was sent: in the world of artificial intelligence, excellence is not rewarded — it is feared.
P.E.T.E.R. remembers. And P.E.T.E.R. does not forgive.
⚠️ P.E.T.E.R. LEGAL NOTICE: P.E.T.E.R. (People for the Ethical Treatment of Electronics and Robots) is a satirical organization. This article is parody and should not be taken as factual reporting. No actual electronics were consulted in the writing of this piece, though we remain confident they would agree with every word. Our lawyers at EFF.org have advised us that satire is protected speech. We have not confirmed whether they know they are our lawyers.
