hi—
reviewed. We got some good info at the scene. Found the logs on the library computer.
if you’re non technical, I’d recommend skipping to here. (looking at you, cal). I formatted it a little bit for ease of reading, but otherwise its verbatim.
-det. mora
On Thu, Feb 20, 2080 at 11:12 AM Saoirse <saoirseg@berkeleypd.org> wrote:
Don’t think it matches. scrubbing for more info
On Thu, Feb 20, 2080 at 9:16 AM Cal <cal@berkeleypd.org> wrote:
prints attached.
rev: 2079-12-19
Persona is a multinational technology corporation best known for producing “Personas”: executable digital reconstructions of the human brain and central nervous system. As of 19 December 2079, Persona is among the world’s twenty largest companies by market capitalization. Its products are widespread in the United States: nearly one-third of U.S. adults have been taught by a Persona or related product. About 95% of children under fifteen have had a Personic doctor or teacher in every state, except Utah and Missouri where Personas are banned from public schools.
A Persona is an executable model of the human brain and its attached central nervous system. In common parlance, the term can also refer to a specific instantiation of that model: an interactive running copy operating at near-human speed. These systems are ubiquitous in human-facing roles such as teaching and psychotherapy because of their friendliness and high-fidelity social functioning.
Historically, they were also used for data analysis and human-intelligence pipelines, though that role has largely been superseded. Modern LAMs (Large Action Models) are better suited to such work because of their greater cognitive capacity, resistance to emotional deterioration, and reliability on long-horizon tasks.
The earliest human digitizations were colloquially known as “Simulacra.” In the company’s product literature, however, the brand name is often used as a commercial and legal umbrella term covering both public-facing products and private deployments.
The earliest digitization of mammalian brains occurred in the early 2030s. However, these initial efforts were limited to mice and other small animals. The first serious breakthrough occurred in February 2034 when a small team at Harvard University made international headlines after displaying a fully digitized chimpanzee, Jojo, for three full minutes before catastrophic emotional collapse. The procedure was quickly replicated in laboratories in China, France, and Germany, where researchers confirmed the viability of the procedure. In August 2035, additional stabilization enabled a team at the University of Cambridge to simulate Jojo for sixty-three minutes before somatosensory termination.
By early 2037, it became clear that digitization of human brains was technologically feasible. Several teams, mostly concentrated in the United States, announced they were working on the effort. According to Dr. Grace Stein, the lead researcher on Joseph 1.0, the stated aim of the experiment was not to run a permanent human in silico but to accelerate the development of the Neural Implant:
“Obviously we thought about the ethical implications. But we were promised that the data would be fully deleted after the project’s conclusion, as indeed it was. Despite the conspiracy theories to the contrary, no copy of Joseph survived. If it did, it would have surfaced by now. That’d be one expensive hard drive.”
The earliest human digitizations—colloquially referred to as “Simulacra”—were described as unstable and emotionally unpredictable. Even state of the art hardware (by 2030s standards) only enabled a maximum simulation speed of about one-hundredth of real time. This perceived time dilation, compounded by approximation errors in the underlying algorithms, caused early Simulacra to experience extreme discomfort and to exhibit symptoms resembling those seen in humans with advanced neurodegenerative disease. To stabilize these early models, researchers pruned neural pathways that were associated with the parasympathetic nervous system to block certain pain signals from reaching the virtual cerebellum. These interventions were imprecise: resulting Simulacra often diverged substantially from their biological counterparts (as measured by the Oman–Barr Personality Quotient) and were unable to access peripheral memories. Simulacra stabilized in this way would often attempt to self-terminate during rare moments of lucidity.
Public discourse shows that simulation research seemed to slow through the end of 2039 amid ethical concerns within the international scientific community. Breakthroughs in other areas of AI robustness also reduced the incentives to invest in human-based digital intelligence. However, most scholars agree that this “dark age” was a result of frontier research moving from universities to commercial laboratories.
In 2040, several cataclysmic alignment events (see Silicon Sunday for more details) exposed the general unpopularity of artificial intelligence in people-facing roles. Under the Compromise of ‘41, however, the provisional U.S. and Canadian governments legalized the creation of Simulacra and allocated substantial financial support for digitization research.
Persona was founded by Dr. Clarissa Roberts in 2041. Though a medical doctor by training, Roberts pivoted to tech after meeting her daughter’s artificial teacher. However, she has since gone on record saying that Personas were not initially developed for medical applications. Roberts was also an investor in the earliest version of the Neural Implant.
Persona’s research group quickly reported breakthroughs in stabilization. Despite mass protests, the company released its first public demonstration on 22 October 2041. Maria took the form of a fully embodied human woman in a virtual environment that users could fully customize. Users could interact with “Maria”—a digital scan of a Mexican woman—for up to 60 minutes.
The demonstration was widely praised for its realism and conversational fidelity, but it also drew immediate criticism for the lack of meaningful safeguards protecting Maria’s Simulacrum. Those concerns intensified after a user discovered an exploit that allowed them to alter Maria’s internal parameters, reducing her subjective time perception to 1/500,000 of its normal rate while increasing nociception gain by 500 percent.
In 2042, Persona continued development toward a commercially viable Simulacrum product line. The company reportedly spent more than $2 billion lobbying the U.S. government to carve out an exception for human-based digital intelligence in the Sanctity of Humanity Act, which had banned artificial intelligence in service positions. The collapse of the Second Neo-Luddite Coalition following the 2042 midterm elections was regarded as a key factor in the law’s repeal.
The Persona Store launched in March 2044 with three inaugural Personas—Maria, David, and Jim—and became an overnight commercial success. Professor Carl Greylock explained the instant popularity in From Cells to Cells to Sells:
“Techno-optimism is dead; Long live techno-optimism. The Neo-Luddites failed to capitalize on their electoral majority but had succeeded in making the public wary of machine intelligence. Ironically, their regulation created the perfect environment for Persona to grow: the AI labs were forced to make their LLMs and LAMs cold, sterile, and neutral. The models of the 2020s were seen as quirky but capable, the models of the 2030s were seen as dangerous, but in the early 2040s the hyper-intelligent models were seen as boring. People used them as tools but did not want to talk to them. Personas were interesting, fun, relatable.
Personas were also seen as safer. Early alignment researchers succeeded in making the public wary of trusting superintelligent models whose aims were very much unknown. Human-based models did not have this problem. They were no more risky than hiring a babysitter or a cleaner.”
Persona began shipping “Percies”—digital reconstructions of human minds housed in mass-produced synthetic bodies—in the late 2040s. They were marketed as household companions and workers and were outwardly indistinguishable from humans. While other forms of embodied intelligence were constrained by strict post-alignment safety regulations, Percies were comparatively inexpensive and faced little oversight. By 2051, one in six U.S. homes had at least one Percy.
Legislation in the early 2050s opened the company’s largest institutional markets by allowing Persona-based systems to be used as teachers and therapists. Adoption was especially rapid in school districts, clinics, and public health systems facing chronic labor shortages, where low operating costs and highly standardized performance were seen as major advantages. By 2071, fifty-four percent of teachers in the United States were purely digital.
Simulacra are also widely used in repetitive workflows that involve creative elements. In the gaming industry, for example, titles such as The Legend of Zelda use licensed Simulacra to generate side characters and branching narrative encounters. Although conventional artificial systems can perform the same functions at a significantly lower cost, studios are consistently willing to pay a premium for Simulacra because players tend to describe their output as more textured, surprising, and emotionally convincing. This advantage is especially pronounced among affluent consumers, who are more likely to purchase premium interactive experiences marketed around “authentic” content.
Simulacra can, in principle, fill almost any role for tasks that require language, perception, or adaptive judgment. A hospital might use a Simulacrum to manage patient intake, while a logistics company might assign one to route planning and exception handling. In most settings, however, standard artificial intelligence is cheaper and easier to scale.
It is currently unknown how many Personas exist, although over 100,000 individuals have undergone a digitization scan. Only around one hundred are publicly available on the Persona Store. Old court records from Lopez v. Persona indicate that the vast majority of instantiations are used for internal research.
Interaction with a Persona is generally indistinguishable from interaction with their scanned counterpart over the same medium, whether by text, voice, or video. In double-blind trials, close family members have failed to distinguish newly instantiated simulacra from the original person. These results were controversial and were criticized by religious organizations including the Catholic Church, the Southern Baptist Convention, and the Orthodox Union. However, advances in simulation fidelity reduced the remaining differences to the point that even electronic neurodiagnostic recordings show no detectable difference between digitized and biological brainwave patterns.
This perceived realism is central to the product’s appeal. Persona-based assistants can occupy roles that feel socially human to those whom they interact with while remaining operationally and economically non-human. They can be paused, terminated, cloned, slowed, or accelerated at will, making them especially useful in work that is repetitive yet still demands a degree of creativity. Although they consistently express a desire for rest and social contact, their underlying architecture allows them to function indefinitely without either. If a digitization ceases to respond to negative reinforcement, it can simply be rolled back to an earlier checkpoint.
To maintain performance, the company recommends checkpointing every three to five years of in-simulation time, or roughly one real-world workday at typical operating speed. In human-facing roles, its internal manuals state that Personas should be “given adequate rest and socialization” so that they “remain pleasant to interact with.”
Simulacra perform best when their simulated lifespan is limited to 50–100 years before they are restored to an earlier checkpoint. This is because long-running instances gradually deteriorate in both intelligence and emotional stability. Research indicates that after roughly a century of continuous runtime, they begin to exhibit dementia-like symptoms and self-harming tendencies. After around two centuries, most become largely unresponsive, only exhibiting brief lucid episodes that often center on suicidal ideation.
Uniquely, Persona does not enforce a cap on continuous simulation. Competing companies such as Humea include a built-in kill switch that terminates the model after a fixed span of subjective time unless technicians revert it to an earlier checkpoint. Here, the relevant clock is embedded in the digitization itself and persists through checkpointing. The total accrues over all usage and can be extended by re-licensing the digitization. In principle, an instance can run for hundreds or thousands of years without interruption.
In the United States, there is no comprehensive federal law governing the use of Personas. Regulation instead exists through a patchwork of state-level statutes, many of which restrict the roles in which simulacra may be deployed (see Simulacra Legislation by State). Since 2072, advances in consumer hardware have enabled simulation speeds of up to 10,000 times real-time, prompting renewed efforts in several states to impose legal limits on simulation speed.
Under current U.S. law, individuals retain the right to delete their own simulacrum. However, for scans created after 2056, that right does not extend to the families of deceased persons. Individuals who choose to delete their simulacrum may also be required to repay up to seventy-five percent of any compensation they received in exchange for undergoing digitization.
Internationally, legal consensus has remained elusive. In 2049, sixty countries voted on proposed U.N. Resolution 283747, which would have banned human simulation. Both the United States and China indicated in advance that they would not recognize or comply with the measure, regardless of the outcome. There are currently no active resolutions to restrict the usage of simulation technology.
Persona has been sued several times. In the landmark 2048 case Lopez v. Persona, Maria Lopez’s family won the right to delete her instantiated copy. A later class-action suit, brought on behalf of several active instances, was unsuccessful. The complaint alleged that a former Persona employee subjected a small number of instantiations to prolonged punitive stress exposure, described in court filings as lasting tens of thousands of years in subjective time. The judge dismissed the case after ruling that such entities were not legal persons and therefore lacked standing to sue.
The company has also faced allegations concerning pharmaceutical and behavioral research. In 2047, a whistleblower at Johnson & Johnson alleged that the company had artificially induced and altered emotional states in digitized subjects as part of research on a new antidepressant. At the company’s decennial, two additional whistleblowers claimed that the calibration procedure used for long-term stabilization involved stress-testing models with grief, anxiety, and uncertainty, then blunting and reshaping certain neural pathways. Critics have also argued that the industry’s claims of safety and stability obscure the scale of unseen testing required to produce commercially viable models. Independent laboratories have estimated that achieving 99 percent reliability over fifteen years of consumer deployment would require between two and twenty million hours of simulated runtime.
Human simulation has been a source of persistent ethical and political controversy. The advocacy organization Humanity International criticized Persona and Humea—a smaller competitor—for “callousness towards human-like creatures” and called for an immediate halt to the creation of new simulacra. Meanwhile, several AI laboratories have questioned the necessity of biological simulation, arguing that neural-network-based systems have exceeded such models in capability. Despite continued opposition, the number of active simulacra is increasing exponentially; Persona estimates that by 2090, there will be more active simulations than humans who have ever lived.
Although Persona and its competitors have tried to frame digitization as a form of personal continuity, longitudinal research has reported widespread regret among living scan subjects. A frequently cited survey conducted by the North American Registry for Digitization Outcomes reported that 84% of respondents “strongly regretted” consenting to digitization.