Honorable mention / 5 May 2026, Goodfire-led research team, which trained and analyzed the experimental model; no separate deployment operator was identified.

Goodfire researchers decompose language model parameters using adversarial ablations

The model retained coherent output even under twenty steps of adversarial masking designed to remove non-essential subcomponents.

Honorable mention, not a full case file. No two named parties were found who disagree about the nature of the system.

The question this event raisesDoes the language of intention (wanted, decided, collaborated) describe what AI agents do, or mislead?

Edits show VPD identifies real computational machinery

“Nevertheless, the ability to make precise edits like this demonstrates that VPD identifies real computational machinery within models. And it does so well enough that we can begin to control it directly.”
Goodfire research team, ran the system

goodfire.com, 5 May 2026, archived copy

These readings say something about what the system is, not only about what it did. A language model scored each reading on that, and the ones above a fixed cutoff appear here. The site takes no position.

What happened

Researchers introduced adversarial parameter decomposition to split language model parameters into simple subcomponents. On average, each test datapoint used 205 subcomponents per sequence position. The team manually rewrote parameter subcomponents involved in emoticon predictions.

First-hand sources

Also on the record

These readings describe what happened or judge how dangerous it was. Each quote appears word for word on the linked page.

  • Lee Sharkeyscientist

    “Our decomposition natively handles attention (!) and behaves less like a lookup table and more like a generalizing algorithm.”

    x.com, 5 May 2026, archived copy

What would settle it

  • Tests checking whether direct subcomponent edits reliably alter specific model behaviors out of distribution would separate algorithmic control from fitting artifacts.
  • A comparison of model outputs under larger numbers of adversarial steps would distinguish genuine minimal circuits from brittle approximations.

About agency

One view holds that these isolated subcomponents reflect genuine algorithmic mechanisms that allow direct manual control. Another view treats them as a mathematical decomposition that approximates general function execution without proving underlying intent.

About consciousness

This event is evidence about what the system did. It does not establish or rule out that the system experiences anything. The readings on this page disagree about what the behaviour shows.