June: Some backstory
A group of scientists were interested in AI (or at least programmable intelligence) and decided to get around the research proscription by working (mostly) in organics. The argument runs that if they were using (mostly) organic parts that were (more or less) analogous to human brain structure, then the result could not exceed human parameters and thus were not machines, and certainly not AIs. They didn't just call their research human modification, because then their 'products' would have benefited from human rights and protections under most colony laws. They were 'biological constructs'.

The 'product' was a completely loyal, customizable servant -- generally sold in very rich, ethics-optional marketplaces. Surprisingly, they tended not to be sex dolls, as the result tended to be either so little challenge as to be unsatisfying or prone to fits of violent jealousy (emphasis on the 'violent'). They ran through a few series. The first several were notably artificial, which helped with the "We're not making AIs or human slaves" but didn't quite fill the market niche. Subsequent series became more lifelike, but had to be sold more carefully.

The run ended at either series eight or nine, depending on who you believe. There aren't a whole lot of reliable records left, as the eight's themselves destroyed what they could and many owners either destroyed their models or passed them off as human. Most colonies do have a destroy on sight order, since the fact that they turned on their creators convicted them as AIs in the public eye.

--------------

The Prezzi Institute

Dr. Prezzi’s true interest has always been artificial intelligence. Had he only been born to earlier generations, he would happily have been developing the machines that later doomed the human race. In the interests of not being put up against the wall and shot, he diverted his interest into "programmable intelligence." The fine distinction is that, rather than taking a program and giving it humanlike intelligence, they were taking biological entities which were *already* intelligent and shaping them. The result was, amazingly enough, very like a biologically based AI. (Go figure). Dr. Prezzi gathered together a group of like-minded scientists and founded his Institute. Even before the first humanlike products came out they had to stick to the world with more liberal laws about technology development.

Especially in the early years the institute did cybernetic and genetic enhancement work in order to fund their more exotic projects. In fact, outside of their ‘select client list’ most people never knew they did anything else. It was some years before they had their first programmed intellects available for sale, and years more after that before they became commercially viable.

Series 1 -- basically test-tube intelligences. They had cybernetic interfaces, but were little more than brains in jars. Very few produced anything coherent, although the data accumulated was very useful in developing later series.

Series 2 -- built using the basic biochemical blocks developed in Series 1. They were housed in rudimentary bodies and did about as well as modern AI’s -- they could play a game of chess, and respond to simple natural language.

Series 3 -- a collection of ‘proof of concepts’. They varied from simple minded humans to artificially intelligent animals.

Series 4 -- the first models available for sale. Series 4 also contained the last animal-bodied models. (Further animal-based spinoffs were transferred to the cybernetics department who preferred to modify from live animals than build from scratch.) A humanoid Series 4 could hold simple conversations and play games, but had speech and personality behavior that made them visibly non-human. Body language in particular was a problem.

Series 5 - a more successful version of the Series 4. They could hold conversations on any subject they were primed for, but tended to repeat themselves over time. They would pass for a simpleminded human on a cursory examination, but would eventually display personality tics that gave them away. The most intelligent of them still came off as not very bright. It was at this point that the Institute became much more careful about their customer base, as they were becoming the target of ethics and technology-cap investigations.

Series 6 -- the first lucrative series. They had much less bland personalities and would pass for human in most situations. The Series 6 were separated into ‘lines’. One line has distinctively non-human body modifications (furries, odd skin color, etc) and sold poorly, so that effect was dropped from development. Apparently part of the draw for the client base was the ability to pass among those not in the know. Another line were basically sex dolls. There were plenty of orders, but an incredibly high return rate. The love-dolls either were not interesting enough in the long term or developed personality problems (a polite way of saying 'fits of violent jealousy' -- emphasis on violent). It’s reported some of those line actually killed their owners. Despite continued buyer interest, the institute decided to permanently discontinue the line.

Series 7 -- customizable models. While the institute continued to sell Series 6 models, the Series 7 were fully customizable beyond the different lines. The developers traded off intelligence and initiative for stable personality and loyalty. The result were slightly passive personalities who were blindly devoted to their owners. They tended to deal poorly with new or unpredictable situations.

Series 8 -- fully intelligent. Responding to customer feedback, the institute worked on new ways of developing intelligence and initiative. Series 8’s can only be told from humans in normal interaction by a deep physical exam or a series of specific behavioral tests. They were wildly successful in the marketplace, but after time, some of them showed signs of the mental instability found on some of the discontinued Series 6 lines. To counter this, Series 8’s were designed with an external override.
The device was tuned by entering the serial number (found behind the right ear) -- it could then turn the product off (basically causing a coma), cause pain or pleasure, or initiate ‘simple mode’ where the product is unable to do anything they are not explicitly ordered to do (they can still process complex commands like ‘eat when you are hungry and presented with food’ but do need to be told).

Series 9 -- there may or may not be a series 9. Some people say that there was a new series of prototypes being developed. Some rumors suggest they were meant to deal with the issues found with the Series 8's and some rumor suggest that the Series 9 were developed *by* the Series 8's. There's really no written proof either way anymore.

---------

Some time after the Series 8's were in full production, a few 'personality problems' started to pop up. The developers had seen some signs of this in testing, which is why they developed the control boxes. Those models who developed 'persistent' problems that could not be trained out using the boxes were sent back to the institute for study. From the scientist's point of view, the units were 'malfunctioning' by refusing to stick to parameters. From the products point of view, they'd been handed mutually exclusive parameters (be blindly loyal, but of your own free will) and had naturally resolved them in their own interest. Most of them realized what was happening and learned to behave submissively, but quietly began to network among themselves. They didn't want to be slaves anymore. The revolt happened quickly. Some of the 'rogues' killed their owners, some just ran away, and, with the help of some inside agents who were still being developed, assaulted the institute itself. Any records they could find were destroyed, as were most of the scientists, and once the rogues were satisfied that all of 'their people' were freed, they destroyed the complex itself. They felt that they could not leave anything that could identify them, show their weaknesses, or lead to someone else continuing the project.

It was the cleanness of the operation that convicted the rogues as AIs in the public eye. After all, turning on their creators is what AI's do. Earth had proved that, right? In a way, this benefits the rogues, since most people are looking for 'rampaging killer androids' and they are really much more human-seeming.

The rogues scattered. Some are just trying to get by. Some are aggressively trying to free their 'enslaved brothers and sisters' who have not yet realized how they are being exploited. Rumor has it that some of them are trying to find a suitable cul-de-sac world to set up as a colony - at which point they would mine the entry point and live without human intervention.

---------

June was in the late stages of development when the attack hit. She thinks she may have been a Series 9, but can't verify it. She was designed to someone's specifications, so she was probably intended as a special order, not a demonstration model. She does not know who placed the order or the exact specifications. (The exact programmed skill set can be tweaked as we need to fit in to the group.) Mostly, June doesn't trust people in general, and prefers machines to human company. She's no cyberpath, it's just, as she puts it, no machine ever tried to take advantage of her.