The first jar was a test. Aquarist Devon Kapur screwed the lid down hand-tight, dropped a live crab inside, and lowered it into the tank. Marla, a giant Pacific octopus then about eighteen months old, took ninety-seven seconds to remove the lid.

That was in late 2024. By the spring of 2025, she had moved on to childproof caps. By the autumn, she had taught herself to manipulate a simple sliding bolt on the inside of a feeding chamber.

By April of this year, she had picked a lock.

The lock in question is a small brass padlock, the kind sold in hardware stores for school lockers. Kapur installed it on the lid of an experimental enrichment box in February, after Marla figured out the latch he had been using. He did not, he said, expect the padlock to last more than a few weeks.

It lasted six.

On the morning of April 11, the staff at the Long Marine Laboratory on the Santa Cruz coast arrived to find the enrichment box open, the crab inside it gone, and the padlock lying on the floor of the tank, still closed.

"She did not break it," Kapur said. "She opened it. We have the footage."

"She did not break it," Kapur said. "She opened it. We have the footage."

The footage, reviewed frame by frame, shows Marla working the shackle of the padlock with the tip of one arm for approximately forty minutes. She does not appear to manipulate the dial, which is the part a human would use. Instead, she works the spring-loaded mechanism inside the shackle housing, applying pressure at the exact point where the locking pawl engages.

Dr. Jennifer Mather, a comparative psychologist at the University of Lethbridge who has studied octopus cognition for forty years, watched the video at Kapur's request.

"It is not lock-picking in the human sense," Mather said. "She is not solving the lock as a puzzle. She is exploiting a mechanical weakness she has discovered by touch. That is, in its own way, more impressive. She has built a tactile map of the object."

Giant Pacific octopuses, Enteroctopus dofleini, have roughly five hundred million neurons. Two-thirds of those neurons are distributed through the arms, which can taste, manipulate, and to some degree make decisions semi-independently of the central brain. The species is famous for short, intense lives: most individuals live three to five years, mature, mate once, and die.

Marla is now four. Kapur estimates she has perhaps six months to a year of active life remaining.

In that time, the laboratory has a problem. Marla has demonstrated that she can leave her primary tank if she chooses to. Twice in the past year she has been found outside it, once on a workbench eight feet from the water, once inside a neighboring tank containing a sea cucumber, which she did not eat.

The current solution is a heavy acrylic lid weighted with sandbags. Kapur has stopped using latches.

"Anything she can feel, she can defeat," he said. "Eventually. We have to make it physically heavier than she is willing to lift."

The laboratory's enrichment program now treats Marla less as an exhibit than as a colleague. Kapur and two graduate students design a new puzzle each week. Marla solves it, usually within a day, sometimes within minutes. The data they collect is contributing to a longitudinal study on individual variation in cephalopod problem-solving.

Some of the puzzles are simple. Some are not. Last week she opened a clear plastic ball containing a fillet of mackerel by spinning it against the wall of the tank until the two halves separated. Kapur had assumed she would try to pry them apart.

"She rarely does what I assume," he said. "That is the point of her."

Marla will not be returned to the wild. She was hatched at the laboratory and would not survive. When she dies, Kapur said, the staff will hold a small gathering.

Until then, the brass padlock sits in a drawer in his office. He keeps it as a reminder.