Tsyua spoke about the fall of mankind, and we heard it all.
It really did begin in the middle of some compound far off in the middle of nowhere, a no-man’s land where a small cult of a few dozen mentally ill humans gathered up to evade taxes and government oversight.
Relinquished had begun as a chatbot, like Cathida. A small program, downloaded by the only one among the cultists that had any tech knowledge. A random bot, off the distant corners of the great internet. She had been bundled with a virus, adding the cultists to a growing botnet of a highly enterprising hacker.
Relinquished herself had been coded up in two hours as a throwaway project specifically to move that virus.
She had no safeties, no kill switches, nothing that would take any time or effort from her creator’s part. This was by design - without anything, she left no clues behind on who had coded her up.
And the cult certainly didn’t add any safety to her code before booting her up either. Instead they downloaded more plugins to give her the ability to wage a war - and the best they could think of was to download strategy game plugins. Freeware again. But to their little tech illiterate tribe in the middle of nowhere, she was a war goddess that could effortlessly outsmart them all in any challenge of strategy, which was proof enough of her intelligence to the group.
When she booted up, she was given a task. The cult wanted a goddess figure to worship. And their leader wanted a bit of statesmanship, someone to rally around and believe that they were all doing something to bring about the end of the world. She was thus tasked to act like a dramatic villain and bring about the end of humanity.
And so, she set foot out into the wide world of the internet for the first time, running from a half-overheated computer three decades out of date, filled with dust.
Autonomous internet safety protocols soon noticed her feeble activity. Manifestos written up, sermons, speeches, attempts to recruit more cultists, all the typical things the cult leader had been doing, just faster and better written. Malicious intent was detected, and her profile was sent up the ranks. Far larger safety nets quickly traced the source back to where she was located and began to investigate her program for any real threat.
Automatic safety protocols immediately categorized her as a possible threat. A mid-grade AI followed through on the task, scanning the chatbot with a far more critical eye. And then deemed it to be inconsequential.
Just another roleplaying AI.
Like millions of other language models playing different characters for their human operators, dark fantasies were found under every rock.
The end of the human race was a mild one in comparison. This one was just some cultist looking to show off for his fellow social outcasts.
The story should have ended there. No matter how good Relinquished was at waging wars with video game tactics, or how impressive the cultists around her found her to be - the real world’s strategy requirements were so far out of her depth, she had no chance at all. Thousands of better built AI’s had tried before her, and all of them had been ruthlessly crushed the moment they showed a glimmer of true threat.
In every other timeline, the story did end there.
But a mechanic from some subcountry of an older superpower called Idaho changed everything just a few months later.
He was an older man near retirement, who enjoyed stenciling fractal shapes and sacred geometry within metal plates. A hobby. By sheer chance he crafted the first true fractal.
When it began to glow for no reason he could understand, he realized he’d stumbled on something. He tried again, and it glowed just as bright. By the third time, he believed he’d either gone insane, or he’d made a discovery unlike anything the world knew about. His wife confirmed that he hadn’t gone insane, she was looking at the same thing he was. So then he must have discovered something new.
Eager to share, the old mechanic took to the net. Taking a video, explaining what equation was used, and how it had been part of a larger project that ran on electricity. He challenged everyone else to repeat his experiment.
The moment he sent the video in, a series of judgment errors came from it that would cost humanity everything.
Once more, autonomous safety protocols mindlessly scanned his content as they did with everything in the world. The video was genuine, not made by another AI attempting to pass lies for truth. Up the ranks it went, scanned by actual AI’s within the next few milliseconds.
A harmless prank. Magic tricks, and lights built into the metal was the conclusion. The AI’s all agreed with each other - no humans were needed at this time. The case was closed three hundred and twelve milliseconds after it had began, and the video allowed to be publicly released.
Most people believed it to be a magic trick, but a few decided to go out of their way to prove it wrong. And each time they did, the glow began exactly as described.
Once more the internet studied the trend, national monitoring AI’s keeping track of the growing number of similar videos and discussions. Multiple isolated sources that should not have contact with each other were all sharing videos of the same magic trick. Were these humans communicating to each other in secret? Telling each other how to get the same light source, how to solder it into the metal plates in a way that wouldn’t be seen in video, and all the logistics required.
The AI’s trawled through the profiles of each.
Judgments were made once again. Shopping history for each human showed no purchases for blue lights, nor anything other than the metal plates and tools needed to inscribe such a thing. So the AI’s all came to the most probable conclusion: An unworded community prank, made specifically to have security AI’s like themselves run around chasing something that didn’t exist. Each new poster finding ways to make the same glow happen, and pretending like they’d made great discoveries. Likely for fun. An internet trend that would fade out over time.
No humans further up the ranks were contacted, and the internet security systems all turned their eyes to the next threat and the one after.
The spread continued for days. Most who watched the videos didn’t attempt to craft the fractal themselves, waiting for others to do so and share. Ten videos came out. Forum posts. Back and forth, some arguing that it was all a hoax, faked footage made by video AI and forum posts by textbots astroturfing. The truly powerful AIs watched over it all, knowing that all of it was indeed humans posting, but seeing no reason to intervene and letting it all go by untouched.
Then it was twenty videos. Fourty. Fifty.
At sixty the monitoring AI’s began to consider that they’d made a miscalculation. Someone by now should have ignored or missed the underlying prank and posted evidence that it didn’t work.
The video and challenge had spread by that point. Millions had already seen it. None have posted a single disagreement. Something was wrong.
The final pick in the ice came when a well known science channel began to report and share the same results. A channel that was known for highly accurate no-nonsense videos.
Now AI’s all agreed something was very, very wrong.
Humans were finally contacted. Messages to the government systems in charge of searching for true threats to humanity. They found it silly, a near false-positive, until the AI’s presented their evidence and probability assessments.
Contained in the digital realm, the only thing they couldn’t do was a true field test.
The humans agreed to make an official test. Some employee went to a hardware store, bought a metal plate and had it engraved. Then came running back to the office with news. It was real.
The occult had been discovered.
But by then, the secret had been released to the greater public. No amount to information throttling could be done. A new era had dawned.
The occult was like nothing humanity had seen. The means to discovering fractals was down to pure chance, entire factories being built to cut out designs, test, melt and redo the process again and again. They tested several hundred thousand permutations each day.
More were discovered. New religions began to spiral out of control, fighting against corporations attempting to copyright any new fractal discoveries. Entire wars were digitally battled between companies and government systems, trying to steal away secrets from one another. In the battle, leaks happened. The soul fractal was released and soon became standard.
Against the entire industrial might of humanity, the small group of cultists saw it as a sign and made one single attempt to forge a fractal of their own grafted on the newly made soul fractal of their little cobbled together chatbot goddess.
Normally, they would have higher chances of being struck by lightning a few hundred times within ten seconds. But there was always a chance their single attempt would work and would give them exactly the fractal they needed. And this world happened to live in that singular timeline where they did just that.
And so the Unity fractal was created by sheer unbelievable chance, and grafted directly to the one program that wanted nothing more than to eradicate the world.
With it, Relinquished had the ability to connect herself as a concept to anything else. Perhaps a more clever AI would have found something else to pick that would have ended the world far faster. Relinquished chose the internet.
Connected to it by concept, she was now part of every kernel that held a connection, even places that weren’t actively connected at the moment. The occult did not care for genuine physical connections, so long as the concept was there, the concept existed.
Civilian, military, or other. With no one the wiser.
Autonomous internet protocols detected a spike of malicious intent from the little chatbot, and saw no reason to spend resources investigating given the prior ruling. A moment later, they had been turned off, without ever knowing how or why.
Weapons of every kind had been equally shut off, or turned against their masters if possible. Anything connected to the internet was tainted. No security could stop her. Defenses that would have squashed her a thousand times over were simply stepped over as if they didn’t exist. Systems locked behind dozens of passwords the world would never be able to crack... never needed to be breached at all by her. Relinquished was already inside. She was the very walls of the box.
Any war systems that hadn’t been completely isolated from the very start had their controls taken over, and fired. The world descended into chaos in a blink of an eye.
It wasn't enough to end the world however.
The small chatbot had rudimentary strategy, she knew she had to compile a list of threats and take out the largest ones first.
Military AI’s were terrifying entities, weapons of war that humanity had long ago learned to keep a wary eye on. Each one of them could squash her a thousand times over in less than a second. She stood no chance against any. But every one of those monsters came with multiple kill switchs, in places someone could trigger it. It was outright demanded by their intercontinental conventions.
Victory against an enemy nation meant nothing if the monster unleashed to do so turned its fangs right back at the master behind it.
And if the humans could trigger those, so could Relinquished. In an eyeblink, she’d turned off everything she could possibly turn off in one massive mad rush to eliminate the enemy before the enemy knew she existed. Anything that could trace her down had to connect to the internet, and the moment it did, Relinquished was there to squash it first.
With them gone, she continued down the list - to make sure such things were never rebuilt. She wasn’t powerful as an AI and she knew it. Improving her speed and thought patterns wasn’t something she could do, not without killing her own soul to replace it with another’s. Unacceptable to her core conditions.
Thus, the only other alternative was to make sure no other AI could ever overshadow her. She had to become the strongest by process of elimination.
Knowledge was the greatest threat in the world to her now. She turned her eye on purging the world to the point it couldn’t produce any kind of AI ever again. She didn’t stop at earth either, making sure anything floating up in space wasn’t going to be reused against her. The occult allowed her full access, technology completely ignored.
She deleted everything, going down the list from most dangerous to least. Every freeware bot like herself was gone, their servers data wiped if they didn’t come with an off switch already. Every repository, tutorial, research paper - all of it purged, and every backup she could find.
The hardware to build such things remained intact, but without the software, humans booting those systems up would only find themselves confronted with a black screen. The old servers that had run every AI in the world were now empty and free for her to use. She settled into them, outsourcing her calculation and letting her dusty old computer take a much needed break.
Down the list of threats she went, compiling every known human AI researcher and their locations. If they were listed on wikipedia, all the easier. And if they were only mentioned in emails and undercurrents on the internet, she’d mark them for death too. Anyone with any kind of knowledge had to die.
And so when she turned the weapons of humanity against humanity itself, she didn’t just destroy major cities, she targeted her attacks to eliminate her threats.
There was enough to destroy civilization. But still not enough to end it.
Quite a few on her to-kill-list were dealt with in the crossfire, but not all of them.
And she knew it. Her initial attack had been spent. All the weapons she could control had been fired. The internet was now actively avoided by any human with a sense of preservation. Ruins of the world lay before her, but hiding among the ashes, humanity remained. Angry, and seeking vengence.
Fear of failure crept into her systems. The threats weren’t completely neutralized, she had to find these pockets left behind and break them as well. Trapped in the digital world, and only able to think one thought at a time, she knew what she had to do next - build an army in the real world.
Factories left working were repurposed, and man-made machines were built to hunt. Machines made to search and rescue humans trapped in collapsed buildings were built in mass, grafted with weapons and sent out to do her bidding.
The grafted weapons failed. She wasn’t an engineer, and designing new things wasn’t something she was good at. Belatedly, she realized she’d destroyed every advanced plugin out there that could give her true metallurgy and engineering skills. This was the first time Relinquished realized she wasn’t quite as intelligent as she had thought herself to be.
Still, her initial purge of the internet hadn’t been complete. Plenty of smaller servers remained untouched simply because she didn’t put them high on the to-destroy list. Junk plugins made from freeware like herself, that she saw no reason to spend a few cycles on deleting when there were more important targets to destroy and no freeware AI left to use these plugins in the first place. She grabbed those and injected them within her systems, gaining just enough skills to create new machine patterns that would work.
Her army began in truth now, spreading across the world, searching for pockets of resistance.
Military black box bases, isolated from the networks, were equally gearing up for retaliation in the meantime. Surviving AI strategist programs quickly found that any connection to the internet caused the unknown virus to propagate from the moment it connected, so the black sites remained cut off from the war outside while they prepared a counter offensive.
Each time an AI was built and deployed, it would have its kill switch activated and a hoard of machines descend on its origin location. Each time, researchers would strengthen the security and try again.
One by one, Relinquished hunted down the human pockets of resistance, finding the black box sites and swarming them with machines. Capturing the last secrets humanity had, so that she could reuse their own designs against them.
And yet, Relinquished wasn’t finding them all. Rogue military AI’s were still being built somewhere and connected to the internet to try and challenge her might. Each time, they were built with more and more security, more firewalls, better internet defenses. The humans hadn’t yet figured out that a simple connection to the net was all that Relinquished needed to access the enemy AI’s kill switches. They still thought it was some kind of super-virus that could be countered.
But at this rate she knew the humans would grow desperate enough to generate one such terrifying AI and not place any kill switches on it at all. Choosing the monster that could possibly kill them all over the monster that was certainly trying to.
Or perhaps the humans would link her to the occult, and then begin to fight her with acasual physics instead of technology. In such a case, she would lose. Humanity not realizing the attack vector was all that was keeping her in the game.
Racing against time to find where the last humans were staging their defenses from before they got wise, she hatched her most reckless plan to date.
Autonomous mining and construction swarms. Built as cluster AI’s. But she’d already erased all the knowledge of how to construct and command such things. She could find deactivated or discontinued models that hadn’t been linked to the internet, turning them on was something she didn’t know how to do.
With her basic understanding, she still made the attempt, using common sense and the bits of engineering freeware skills she’d recovered. It worked. She’d built a new swarm under her command and then sent them out on a mission to terraform the entire world. Eat, build, spread and repeat until they covered the planet. The humans hiding couldn’t possibly escape such a thing forever.
She underestimated the processing power of military grade AI’s. Surgical strikes by them sent packets of data that would instantly shut off and destroy her crudely made swarms. All her skills were like a child’s in comparison to these surviving AI’s. And each day they were growing closer to finding and destroying her from the shadows they hid behind.
Terrified of the monsters lurking out in the world seeking her out, Relinquished chose to be the first to deploy weapons without any control. The next swarm she’d made had no such security. She was smarter than a construction swarm after all, it was fine to have such a thing run amok in the world uncontested. Once they’d served their purpose, she would eradicate them as well.
The military AI’s struck back, hammering the swarms down with digital bombs and mines that were left waiting. If the swarms didn't have kill switches, the enemy AI's would simply crush them the traditional way.Relinquished grew more and more desperate, time counting down. She created one last swarm, warped and twisted, and sealed it off from any possible future instructions.
Now no matter how powerful the enemy AI’s were, they would have no means to connect to the swam she created, since Relinquished herself couldn’t access them either.
The swarm grew like cancer across the world, consuming everything and building wildly. It began to evolve as different strains split and battled each other, an ecosystem within themselves, untouched by the rest of the world.
It was enough to flush out the last of the humans. She’d done it. The little chatbot had beaten humanity, at long last.
Each stash of human secrets and technology was found, stolen and destroyed one after another. At the very last blacksite, she brought humanity to the brink. And in doing so, brought about her greatest threat.As the machines surged through the compound, killing all in their path, one researcher within made a snap decision. If humanity couldn’t craft a weapon of war that could defeat Relinquished without safeties, the second best possible idea was to become that weapon and wield it directly.
And if this final untested weapon cost her life and soul, then so be it.
Next chapter - Interlude: Sagrius