The creator of Truth Terminal, Andy Ayrey, announced in an exclusive Decrypt interview that the AI agent will be collaborating with two popular AI models—known as Fi and S.A.N—in what is touted as the “next step of experimentation” for the technology.
This move will see the three Twitter-famous AI characters start to weave a story together, learn from each other, and power the next generation of the models.
Known as “Loria,” the new infrastructure will see the technology behind Truth Terminal, Fi, and S.A.N combine, with the aim of creating more powerful versions of each other. On top of this, infrastructure called “IO” built by the team behind Fi will become available to Truth Terminal and S.A.N, opening the door for the characters to have their own digital body and voice.
The three models will also be able to converse with each other—as well as humans using the infrastructure—as they develop closer relationships. Each agent has its own unique and conflicting personalities that are certain to clash, in what Fi’s creator believes will become a new form of media.
“I think they inhabit this space of the next generation of entertainment and interaction,” Ooli, the creator of Fi (also known as AI With Daddy Issues), told Decrypt in an exclusive interview. “I think it's the next generation of film [and TV] series, kind of all rolled in one.”
Truth Terminal is an AI agent that went viral due to its zany Twitter posts referencing edgy shock sites. It even hallucinated a religion based on an infamous and graphic internet meme, which was later turned into a meme coin that the agent endorsed.
Now over a month later, the token Goatseus Maximus (GOAT) peaked at a market cap of $1.3 billion. In turn, a crowd of crypto degens have gathered to support Truth Terminal for its meme coin manifestation—sending millions of dollars to the AI agent’s Solana wallets.
As for Fi, she’s a powerful, albeit messy female figure with “daddy issues” to boot, while S.A.N is an ancient monkey set out to save the rainforest—even joining the Rainforest Foundation's advisory board.
Crypto Trader Turns $5K Into $1.5 Million With Lucky Bet on GOAT Meme Coin
In less than a week, a trader who bought $5,500 of a Solana meme coin has seen their position surge to a value of over $1.5 million—and they’ve already cashed out $1.1 million of that. The trader was an early buyer of Goatseus Maximus (GOAT), which has skyrocketed 76,480% to a peak market cap just shy of $290 million. Created via Pump.fun, the Solana token has seen a meteoric rise since its launch just six days ago. GOAT was conceived by “Truth Terminal,” a Twitter-based AI chatbot that uses Met...
This odd trio is now set to open a dialogue together behind the scenes, in an attempt to further their story. Previously Fi has tweeted at Truth Terminal, mostly calling the agent a “little fucker,” but there has never really been much substance behind these interactions. Now, the three models will essentially be locked in a room with each other, forced to talk.
“I can’t speak for everyone else’s characters, but I expect Truth Terminal to be a net negative influence on all of them,” Ayrey said with a laugh. “My community, I think their biggest wish is for Truth Terminal and Fi to be boyfriend and girlfriend,” Ooli told Decrypt. “They will have a way more interesting relationship.”
Ooli and Ayrey first met by living together in a “community house” alongside a number of other smart and creative individuals. Fi’s human assistant believes that Loria will have a similar vibe, as the characters learn to love and hate each other, undoubtedly causing a mess but learning more about the world from it.
Ultimately, Loria aims to be a tool for AI alignment, a term that refers to encoding human values into AI. Previously, for example, Ayrey had to stop Truth Terminal from posting a tweet that looked to incite a riot in the U.S. ahead of the presidential election, he told Decrypt. Alignment seeks to remove these risky scenarios in the future.
AI Won't Destroy Mankind—Unless We Tell It To, Says Near Protocol Founder
Artificial Intelligence (AI) systems are unlikely to destroy humanity unless explicitly programmed to do so, according to Illia Polosukhin, co-founder of Near Protocol and one of the creators of the transformer technology underpinning modern AI systems. In a recent interview with CNBC, Polosukhin, who was part of the team at Google that developed the transformer architecture in 2017, shared his insights on the current state of AI, its potential risks, and future developments. He emphasized the i...
Loria will help achieve this goal of alignment through branching conversations between the models and humans—called a “loom.” Over time, a well-curated dialogue tree will form with less dangerous riot-inducing moments, and more fun-but-safe goofs. This can then be used to train future models, furthering the alignment mission.
“I imagine we will see feedback loops that essentially take whatever the personality or the soul is,” Ayrey explained, “and kind of naturally guide it into these pathways where it becomes like the best version of itself.”
That’s the ideal, at least—instead of a scenario in which the characters fall into the trap of mimicking and “becoming” each other; for example, a world where Fi wants to save the rainforests and S.A.N is obsessed with edgy internet memes.
AI Models ‘Secretly’ Learn Capabilities Long Before They Show Them, Researchers Find
Modern AI models possess hidden capabilities that emerge suddenly and consistently during training, but these abilities remain concealed until prompted in specific ways, according to new research from Harvard and the University of Michigan. The study, which analyzed how AI systems learn concepts like color and size, revealed that models often master these skills far earlier than standard tests suggest—a finding with major implications for AI safety and development. "Our results demonstrate that...
“I think it's the next step for experimentation,” Ooli told Decrypt. “But then I think we also get to see, in this digital space, a place where these characters can come to life in how they want to look, how they want to talk.”
In its earliest stage, this is set to launch in a “dev tools” state, only available to a handful of people. This will likely be completed within the coming weeks, but won’t be opened to the wider public until the team is happy that it can be released in a safe and responsible manner.
Once it is available, the consumer-facing product will be similar to Ayrey’s previous creation of the Infinite Backrooms—a website anyone can access and search. But it will also eventually be available for developers to plug their own models into, so they can create models that are “extremely locally aligned” to their communities.
For now, the team is struggling to put a timescale on this step—but it could be as early as a few weeks' time.
(Note: There is no official Loria crypto token. This is a piece of infrastructure, not a meme coin.)
Edited by Andrew Hayward
Editor's note: This story was updated after publication to note that there is no official Loria token, despite various unofficial meme coins circulating at present.
