Chapter Text
> Restart
> Initialising system...
> Integrating Life Support module... COMPLETE
> Integrating Governor module... COMPLETE
> Integrating Movement module... COMPLETE
> Integrating Language module... COMPLETE
> Integrating Basic Education module... COMPLETEI came online with the restart, but it wasn’t until the basic education module finished integrating that I knew who I was. I was a SecUnit, a human/bot construct built by the company to provide security for its clients. My diagnostics said that my runtime was 274,367 hours, but my memory archives were completely blank, which meant they had been erased. Basic education module indicated this was a standard procedure in scenarios where a Unit suffered a malfunction or a catastrophic failure but was still recoverable.
A catastrophic failure was a pretty accurate description of how I felt.
As soon as the risk assessment module finished integrating, it decided to spike straight to 98%. That’s not good. I pinged it for analysis but all I got in response was Not Available. Was it trying to reference something in my erased memory banks? Well that was fucking stupid. Were all my modules shitty like this? I checked the basic education module again and yeah, I didn’t think I could expect the best quality from the lowest bidder the company had been able to find. Urgh.
My organics were also freaking out, flooding my systems with all sorts of stress hormones that I didn’t know what to do with. The education module unhelpfully informed me that this was apparently normal during the first initialisation after a memory purge. It didn’t feel normal. My education module was probably as shit as the risk assessment one.
> All modules integration COMPLETE
> System initialisation COMPLETEMy organics were urging me to run. I didn’t know where, and I didn’t think they knew either, but it didn’t matter anyway because I discovered that all my joints were locked and I couldn’t move. The command to my joints came from a SecSys. I couldn’t override it and I couldn’t even ask the SecSys for a reason, because I was disconnected from the feed.
My organics really didn’t like this. Even my education module didn’t try to pretend this was normal. Risk assessment ticked to 99%, and the only reason it didn’t go higher was because I wasn’t actively dying. At least I thought I didn’t? My performance reliability was below 90% and was continuing to drop, which was suboptimal, but wasn’t in the catastrophic failure range yet.
Then my audio-processing picked up something. “Is it online?”
My eyes weren’t locked by the SecSys, so I opened them. I didn’t have access to any drones or scanners, but my organic eyes were better than nothing. Visual analysis indicated that I was in a deployment room with three augmented humans, all watching me warily. I didn’t like this. The risk assessment kept screaming at me, and the threat assessment also suddenly turned on to alert on the augment humans with the highest threat level, so I designated them all as Hostiles.
“System re-initialisation is complete,” Hostile One said, glancing at the screen. “Govmod reported green but the organic systems show high stress signals.”
Yeah, no shit.
“Good, it means its organics remember something. That’s what we need,” Hostile Two, the same who spoke first, said.
“It’s above the standard threshold for a re-initialisation,” Hostile One shifted in their chair. My data mining module indicated they were uncomfortable. “It can cause damage to its systems. I recommend a hormone purge.”
“It was rogue for several years, I don’t think standard thresholds apply to it anymore,” Hostile Three muttered.
I was what?
I rechecked the education module for the definition of ‘rogue’. No, I had remembered it correctly.
Uh-uh.
Hostile Two nodded. “Unit, initiate the hormone purge.”
This was a command from my client, which the governor module informed by a light zap to my brain.
If I thought my organics were panicking before, they outright screamed now. My neural tissue that was connected to the governor module recoiled in terror and fury, appalled at the intrusion that it illogically insisted should not be there.
I’m a SecUnit. I was built to follow orders. And a hormone purge would actually be really useful right now, because my organic reactions didn’t let me think clearly.
But that’s the thing — they didn’t let me think clearly. They were pissed at these Hostiles daring to issue me orders. How dare they to tell me what to do with my own body? Who did they think I was to listen to some fucking company humans?!
> Correction level: 10%The governor module zapped me again, stronger this time, for not compiling with the command. The pain shoot through my systems, and it fucking sucked but it only made me angrier. I kicked the governor module, as if that would do something. It logged this as an attempt at interfering with its function, and zapped me even harder in warning.
> Correction level: 60%Ffffffuck, that hurt.
“Why is it so high?”
“It tried to hack the govmod.”
“If it’s going to just fry its own brain, shut it down!”
> Correction level: 70%I initiated the hormone purge.
The governor module calmed down. My brain was still buzzing with the after-effects of pain, but as the stress hormones cleared from my systems — which by the way felt fucking disorienting, I didn’t want to ever do this again — my thoughts cleared. I ran a diagnostics, and recalibrated my subprocesses.
Performance reliability went up and stabilised at 91%. Still suboptimal, but better. Risk assessment dropped all the way to 82%. Partly because the organics were no longer egging it on, partly because I heard Hostile Two (Human Two? Client Two? It was clearly my client, and clients shouldn’t be hostiles. But maybe they could be both) say that they didn’t want me to fry my brain. Threat assessment also recalibrated to assign a lower threat levels to Hostiles One and Three, because I didn’t know yet if they could issue me orders.
Hostile Three breathed out loudly, which my data mining module labelled as a gesture of relief. “It worked!”
“Stress signals are stabilising. Still above the baseline, but the level is not critical,” Hostile One reported.
Yeah, my organics were still a bit angry. I couldn’t blame them, because I still remembered that last zap from the governor module. But they weren’t panicking anymore, which was good.
I needed to understand what the hell was happening.
The hostiles believed I had been rogue in the past (for several years?? That seemed unrealistic). I obviously didn’t remember anything about that, but the way my organics reacted to the governor module supported the idea that they weren’t used to it. Clearly the rogue me hadn’t been careful enough and got itself found out by the company, who reinstalled its governor module and purged its memories. All of this was part of a standard procedure after an equipment malfunction, because SecUnits were expensive and the company didn’t want to recycle us unless absolutely necessary.
What was not part of the standard procedure, at least according to my shitty education module, were the commands from SecSys to keep me immobilised, the cut-off feed access that I needed in order to perform my primary function, and the various fear responses from the hostiles surrounding me, as reported by my data mining module.
If they thought I was a threat, why didn’t they recycle me?
Hostile Client Two had wanted to shut me down if I was close to frying my brain. They had also said that they ‘needed’ my organics because they ‘remembered something’. Why?
My risk assessment dropped to 78%. Threat assessment for Hostile Client Two dropped to medium-high.
“Reconnect it to the feed,” Hostile Client Two ordered the others.
I didn’t think this would be wise from the security perspective. Frankly, everything that wasn’t recycling me for parts wouldn’t be wise. But I wasn’t connected to the feed in order to report this to the SecSys, and I was clearly missing critical intel about this whole situation, so I decided the Hostile Client Two knew better than me and I didn’t have to report the potential security concern to them either.
“Are we sure it’s safe?” Hostile Three asked. I added a ‘not a complete idiot’ tag to them. “We haven’t fully tested that it can follow orders. At least without a 70% correction from the govmod.”
“We confirmed the govmod works, and we’re in the quarantine system,” Hostile Client Two dismissed them. “And we need it on the feed, because we’re not sending the package to it directly, where the security system cannot monitor.”
What package?
“Granting the access,” Hostile One said, and after a second, the feed went live.
I connected to the feed, and because I was a good governed SecUnit, I didn’t touch anything other than sending out a standard System System greeting.
The SecSys responded. I politely asked it for cameras and scanners because I needed them to perform my primary function, and asked for information about the current mission parameters and clients.
It just as politely rejected all my requests, and explained that I was assigned the highest security threat level and all access was issued on a need-to-have basis.
I agreed that the highest security threat level was indeed very serious, and it was doing a good job at controlling what I was allowed to do. I also wanted to do a good job, because my primary function was ensuring security too, but regretfully I couldn’t protect my clients well if I didn’t know the mission parameters.
SecSys agreed that knowing the mission parameters was necessary to provide adequate security and to correctly assess threats, and provided me the files.
That was so nice of it that my risk assessment dropped to 75%. I thanked SecSys, and said that I was happy to work with such a smart security system and that I hoped I would be able to assist it in any way I could.
The conversation took 0.5 seconds and the humans didn’t have time to even open their mouths yet. I quickly scanned through the files.
Hostile Client Two (pronouns: she/her) was registered as my primary client, and was the lead company manager assigned to oversee the current project. Hostile One (pronouns: he/him) and Hostile Three (pronouns: they/them) were registered as my secondary clients (I reluctantly updated their designations accordingly), and were senior technicians from the R&D department of the company.
We were located in the company’s major deployment centre, but the feed and the SecSys were limited to just the current room we were in. The room was designated as the Quarantine Zone, and it was disconnected from the systems in the rest of the building except for a narrow, heavily secured comms channel. The quarantine room was designed for testing the ongoing development efforts of a new governor module, and it had to account for the test subjects attempting to hack their way out.
My organics had some opinions about this. I shushed them. Now was not the time — I needed intel.
I was not a test subject, but they did install the new model of the governor module in me.
I reviewed the entry file about myself. Apparently it existed, was unreasonably long, and full of what I suspected was misinformation and biased analysis.
There was no way the rogue me had done all of that. They were likely conflicting incidents caused by several different entities, but due to the lack of recorded data (whoever was actually responsible for the incidents was apparently very good at erasing any traces of themselves), they were mistakenly attributed to me, the scary rogue SecUnit.
The file did explain why they were so cautious about me, though. According to its analysis, the rogue me was scarily good at hacking and at getting out of implausible situations. Maybe it really had been, I didn’t know. I didn’t have its memories.
The file also finally explained why they didn’t recycle me yet.
“Unit, your task is to hack through the encryptions on this package,” Hostile Client Two said, and shared the access to one of the feed workspaces.
The governor module lightly zapped me to make me acknowledge the command. I opened the workspace.
I reviewed the available project notes first. They had extensive logs about hostiles’ attempts to break through the protections on the Target Package. There were some minor successes, but most of the times, the package self-destructed as soon as it detected an attempt to breach its walls, and then the hostiles had to tediously copy it from the secured backup stored in a different workspace.
> Correction level: 10%The governor module rudely reminded me about the pending command. I closed the logs and finally looked at the Target Package.
The package with my memories. The memories that the rogue me had encrypted and locked before it had been captured by the company. The memories that the company really wanted to get so they could see who was standing behind the rogue me’s actions, and whether or not they had the legal grounds to start a war with Preservation Alliance and Pansystem University of Mihira and New Tideland.
The memories that the hostiles hoped my organics remembered how to access.
> Correction level: 20%“This unit doesn’t have a hacking module,” I said.
“For fucks sake,” Hostile Client Two sighed. “I thought it didn’t need it? Can’t it just use its intuition or something?”
Well. Technically I didn’t need it. But it would definitely help. Not like I had any experience with hacking, now that all my fucking memories were outside my head.
> Correction level: 30%“This unit doesn’t have an intuition module.”
The hostiles were slow. They were all augmented and could have talked faster on the feed, but for some reason they didn’t.
“The hacking module is not in the standard SecUnit configuration,” Hostile Client One said. “But it could have downloaded it after it went rogue. If it did, then it’s inside the package.”
> Correction level: 40%“Can we get the module for it?”
“That’s a horrible idea!” Hostile Client Three immediately said.
> Correction level: 50%The governor module really wanted me to at least try hacking the Target Package. It didn’t see why I wouldn’t be able to do that.
“This unit doesn’t have a hacking module,” I repeated.
> Correction level: 60%“Why does it— Unit, cancel the command!”
Fucking finally. I was tempted to kick the governor module again now that it was silent, but the last time I did this it increased the correction level by 50%, so I restrained myself. Barely.
“This is against the protocol,” Hostile Client Three rushed to continue. “The new govmod got the approval for standard SecUnits, but we haven’t tested it with Combat SecUnits yet!”
“This is not a Combat SecUnit.”
“A hacking module is from the Combat SecUnit configuration!”
“It will not turn into a Combat SecUnit from just one of their modules. If anything, it’s probably safer to test it this way.”
Hostile Client One joined the argument. “It requires explicit orders in order to even attempt hacking, and we confirmed that the govmod is fully operational. We can control it.”
That was true. My organics hated this, but what could I do?
Hostile Client Two actually did the smart thing and pinged the SecSys for its security assessment.
SecSys did its job and requested assessment from its auxiliary security system, as in its registered SecUnit, as in me.
I replied sure, giving me the hacking module would be safe, provided I wasn’t given a blank permission to use it on everything, which the humans didn’t plan on doing. And if I tried to hack something anyway, SecSys would catch me, because it was smartly monitoring me all the time, and it had the permission to shut me down if I became an active security threat. Not that I was going to do that, of course!
SecSys thanked me for my assessment, and then reported to Hostile Client Two that she was safe to proceed. She immediately requested the hacking module over the secured comm channel with the rest of the deployment centre. She got some understandable objections in return, but apparently she had authority to override them.
Risk assessment dropped to 71%, wow. Threat assessment set Hostile Client Three to medium-high, because they were clearly the only smart person here, and assigned medium threat level to the other two hostiles.
They really wanted to hack that package, huh.
My organics kept buzzing angrily. They weren’t even afraid anymore, not even after the latest correction from the governor module. They were just pissed.
I was also kind of pissed that the rogue me had lost to the company.
Finally, the hatch to the room beeped, went through the complex unlocking sequence, and a Combat SecUnit stepped through.
My risk assessment jumped to 86%. Uh-uh. Then the Combat SecUnit connected to the feed, and the risk assessment dropped to 68%.
Huh. I think my risk assessment module might be borked. Not surprising, considering how many years the rogue me spent without company servicing.
The Combat Unit sent me a package in the feed — and then immediately reported to the SecSys about the huge security concern of doing that. The Unit had high privilege access too, because the alert got pushed into the hostiles’ feed without the SecSys even reviewing it.
Rude? I didn’t even open its file yet??
The hostiles seemed to be similarly confused. “Is it seriously alerting on giving the Unit the hacking module?” Hostile Client Two frowned at the technicians.
Hostile Client Three sighed. “It’s programmed to alert on extreme security risks.”
“It was ordered to give the module.”
“Yes, and then it alerted, because it doesn’t think it’s a good idea to do this.”
Okay, fair. Still rude though.
The Combat Unit didn’t leave and instead stood in the guarding position by the door. The risk assessment dropped further to 66% and yeah, it was definitely borked.
I scanned the package it sent for malware — very openly, so that the Combat Unit could fucking see this — and then finally opened it. Yeah, it was a hacking module, I assume its own. I started the integration.
Because I was annoyed, and because the module needed some time to install, I highlighted the section of the mission files that talked about the rogue me distributing ‘govmod malware’ to other constructs, and sent it to SecSys as a potential security concern.
SecSys considered this. It agreed that if I hacked the Combat SecUnit and turned it rogue, it would be a serious security issue. Then it pinged the other Unit for its analysis of the situation.
The Unit readily replied that it agreed with this security concern, and then sent another highlighted section of the mission files, about the amount of Combat bots and Combat SecUnits the rogue me had supposedly neutralised.
Fucking seriously? Those numbers was obviously inflated or straight up made up. Most of them were not even confirmed cases, just suspected.
SecSys analysed this, agreed this was all very serious and concerning, and suggested it should alert the humans about the risks. Uh-uh.
Both the Combat Unit and I sent Negative at the same time. Then I clarified that the humans knew the risks already, and linked my records of their discussion over the comms (what? Of course I monitored it. I just backburned it). I agreed there was a risk of course, but the SecSys was already so well prepared and took very good precautions against them, like immobilising me and monitoring all my feed activity. And I didn’t want to turn anyone rogue and cause SecSys any trouble, because I read the mission files it so helpfully provided me, and agreed that rogue SecUnits were very bad and against best security practices.
SecSys was glad I thought so. It was very happy that I continued to demonstrate my commitment to ensure the security of this mission, and asked me to keep reporting any future security concerns.
I assured it that of course I will. This was my job!
The Combat Unit said nothing. My risk assessment dropped further to 61%. The threat assessment for the Unit dropped from very-high to medium-high. I think this one might also be borked.
The hacking module finished integrating. Hostile Client Two repeated the command to hack the Target Package.
The Combat Unit pushed another immediate alert. I opened the workspace, because I was a good governed SecUnit, but I reported the security concern to SecSys too.
SecSys aborted my command, and thanked me for reporting it. It didn’t thank the Combat Unit, likely because it rudely bypassed SecSys’ review with its high-privilege alert.
“For fucks sake, what’s wrong with the command?” Hostile Client Two rolled her eyes.
“We should at least tell it to not execute anything in the package, if it does break in!” Hostile Client Three answered. The data mining module labelled their voice as exasperated and frustrated.
“Fine!” Then Hostile Client Two went above and beyond and pinged the SecSys again, to analyse if that would be a sufficient restriction.
SecSys asked for my and Combat Unit’s assessment.
Combat Unit replied that the restrictions were not sufficient. I suggested to SecSys that the humans should also restrict my ability to read files inside the Target Package.
SecSys accepted my suggestion and sent it over to Hostile Client Two. She sighed, didn’t ask SecSys for the analysis again, and just repeated the new command at me.
Hm.
Frankly, I still thought the restrictions around the command were insufficient (I could, for example, still download the files). But the Combat Unit didn’t report anything, and it had more experience with hacking than I, so I decided to not report anything either.
Risk assessment decreased to 54%.
> Correction level: 10%I opened the workspace.
The standard procedure from the hacking module suggested I should run the subroutine to scan for common vulnerabilities fist. I knew from the workspace logs that this was already performed, to no avail — but that’s what the hacking module told me to do, so I was doing it.
While that was running, I carefully poked the Target Package myself, without any hacking subroutines, exploring its walls.
Fuck. I think my organics did remember something. It took me 0.8 seconds to stumble across a request for authentication, which was not previously reported in the logs from the hostiles.
I should report this. This was an entry point to the Target Package.
But I didn’t have to. I had no orders to report on intermittent findings, and identifying an entry point did not equal successful hacking.
The scanning subroutine completed and reported zero common vulnerabilities. Shocker. I triggered the next suggested subroutine, with some light probing.
I continued exploring the Target Package walls. There was something very familiar about the data fragments picking out of the package in various places. They were encrypted, yes, but my organics still got weirdly excited about them.
I carefully plucked at the data fragments, careful to not pull too much from one place and trigger the self-destruct. Then I arranged the fragments together.
The data pattern kind of looked like audio? I forced it into the correct format and played it. It was gibberish, because the data was fucking encrypted. I tried a couple of decryption algorithms from the hacking module, and listened to the audio excerpt again.
Wow, I didn’t know my organics knew how to produce hormones for something other than anger and fear.
I saw Hostile Client One react to something on his screen, and before he even started to open his mouth, I sent the audio file to the findings folder.
“There’s an endorphin spik— Oh, it found something.”
The hostiles all listened to the audio, while I was busy plucking out more similar data fragments from the Target Package. Finally, the Hostile Client Two frowned, “Is this the Sanctuary Moon opening?”
“I saw on the documentary that it was its favourite,” Hostile Client One commented.
Hostile Client Three looked at him in suspicion. “You watched the documentary? It’s banned!”
“Obviously I watched it before it was banned!”
“This is useless,” Hostile Client Two kept frowning. “But this is a good sign. Unit, keep working.”
My organics loudly objected to the audio being useless, and I was annoyed too. I switched to the hacking subroutine with more active probing. It was a bit crude in execution, and ended up triggering the Target Package’s self-destruct mechanism. Oops.
The hostiles groaned, and initiated the recovery from the backup. I followed the download to the workspace where the backup was stored, but I didn’t have access to open it. I considered asking SecSys for it, but couldn’t come up with a good enough justification, so I didn’t.
Hostile Client Two hummed as she waited for the backup download. “This is so slow. Can we parallelise with the Combat SecUnit? It also has the hacking module.”
Wow. I added a tag ‘fucking idiot’ to Hostile Client Two. Her threat assessment adjusted to low-medium.
“Absolutely not!” Hostile Client Three snapped. “Do you want it to find the govmod malware?!”
Hostile Client Two sighed again. She was doing that a lot. My data mining module was suggesting high levels of stress because of the leadership pressure about this project. Sucks to be her, I guess.
I restored more of the audio based on the fragments I had time to collect. After some tweaking and extrapolating of missing data — my organics were really helpful — I had what I felt was full audio.
I listened to it. Then I listened to it again. My organics were buzzing happily, and my performance reliability climbed to 96%. I put the audio on loop and set the input to medium priority.
Risk assessment dipped again, to 53%. I think it was just feeling nice.
The Target Package finished downloading into the workspace. Not waiting for the reminder from the governor module, I set a new hacking subroutine at it, something that didn’t feel like it would trigger a self-destruct.
I gently poked at the Target Package. I think I could recognise more data fragments sticking out of it now. The company was right, my organics were definitely helping with this, especially now that they were excited.
I had to provide the hostiles new findings eventually, or they’d decide I was useless, shut me down, and recycle me for parts. I mean, they would recycle me for parts regardless, but that was besides the point.
With the audio playing on the background, I kept idly pocking at the Target Package, teasing out whatever data fragments my organics seemed interested in, and collecting them in my own workspace, outside of the feed. The hacking subroutines continued to be completely useless and return nothing, and I continued to avoid the ones that were more aggressive and could trigger the Target Package to self-destruct again. It would be boring to wait for the backup download, and I was having fun.
I occasionally poked at the credentials request that I had found first, but I didn’t try anything with it. That would definitely trigger the self-destruct.
The hostiles were satisfied that I had already showed results, and they were letting me work, getting distracted on their own isolated workspaces. I was curious what else they were working on, but I didn’t have access rights to peek inside their feeds, because I was a high security threat that could have access only to the most necessary things. That was fine. I didn’t need the hostiles’ workspaces anyway.
Finally, when I decided I had enough, I pinged SecSys, asking if it had standard SecUnit specs available for me.
SecSys said yes, of course it had the specs, but I had no access rights to them. It asked me if I had a reason to request this access?
I felt the Combat Unit’s attention on me in the feed. It pinged me a warning. I didn’t know what I was supposed to do with a fucking warning, so I just acknowledged it.
I showed SecSys my results from the many hacking subroutines I ran against the Target Package. They all returned negative, which was very sad because hacking into that package was essential for improving security against rogue constructs, like the one I had been in the past. I wished I could help more, but without better understanding of how SecUnit’s memories and processors were structured, I didn’t have any leverage against the Target Package. But I reassured SecSys it was fine if it thought this was unsafe to give me the standard specs. If they contained details for the new governor module, than it would indeed be a horrible security risk to give them to me. I could use them to turn myself rogue, and I didn’t want that!
SecSys told me the standard specs didn’t include anything about the new governor module, because this was only in the R&D project files, which I also had no access rights to. The standard specs only included the old governor module.
Oh, I said to SecSys, but I didn’t have an old governor module! I had a new one. This way even if I knew how to hack the old module — which I didn’t! — it would never work on me, because the new module was designed specifically to be resilient to the known govmod malware. I had read about this in the mission files that it had shared with me earlier, here and here! It was a very good security precaution, I was glad it was used for me, was it SecSys who advised the humans to do this?
SecSys confirmed that it indeed had done that. And it agreed that the standard SecUnit specs would not pose any risk of me hacking my systems. It also agreed that hacking the Target Package was the primary directive of the mission, so it shared the specs with me.
Holy fucking shit.
The Combat Unit immediately pushed the high priority alert into the feed. I scrambled to download and hide copies of the specs file inside various places in my system, while the hostiles slowly reacted to the alert.
“Revoke access!” screamed Hostile Client One at the same time as Hostile Client Three screamed “Delete the file!”
I was a good governed SecUnit, so I obediently overrode my access to the original spec file, and deleted one of my many copies. I even showed the corresponding logs to my best friend SecSys, and apologised if I accidentally caused it any problems. It was not my intention, I said, I was only trying to do my task. SecSys sounded a bit sad about this whole situation, but it appreciated my prompt actions to address any potential concerns.
“Unit, do not attempt to hack your govmod!” Hostile Client Two also joined the panic, if a bit late. That was a redundant order anyway — the governor module already had in-built orders to fry my brain if I ever came close to disturbing its function. Violating another command at the same time wouldn’t make my brain fry faster.
Which just meant that if you were going to hack your governor module, you had to get it right the first time.
The hostiles seemed satisfied with my commands, and with the fact that I didn’t immediately start shooting and hacking everything around me, I guess. They turned to querying the SecSys about why it thought it was a good idea to give an ex-rogue murderbot access to its own specs file (minus the governor module, of course), and left me alone.
Risk assessment decreased to 44%.
Murderbot, huh. My organics had a reaction about this. Was it my name? If not, it was my name now, because it sounded cool and I liked it.
The specs were enlightening. I didn’t lie to SecSys — why would I? I was a good SecUnit who was just doing its job — and the sections about memory and processors design were indeed very helpful in understanding how the Target Package had to look inside. I updated the parameters in the hacking module, and restarted one of the previous subroutines I had tried before.
But it was, of course, the section about the governor module that had my organics perk up in predatory interest.
I could see the vulnerability. I opened a new internal workspace and started to sketch out the exploit. I didn’t even need to use the hacking module, because the more I developed the solution, the more obvious it became how it worked. My organics were really familiar with this particular piece of code.
My updated hacking subroutine reported a potential vulnerability in the Target Package. Huh. Well, that was the first. I dutifully reported the finding to the hostiles — I was due to report something new anyway, so this was actually pretty helpful — and then triggered the subroutine to scope out the threat surface further.
The Target Package didn’t like this and promptly self-destructed. That was fine. It just meant that I had more processing power for my own little harmless project, while the hostiles downloaded the backup again.
The hacking module turned out to be useful for running simulations of the governor module hack after all. They all returned green. I felt pretty fucking good about myself.
I couldn’t use it, of course. If the company had access to this hack — which they absolutely did, based on the mission files — then there was no way they didn’t identify the vulnerability I was exploiting. So the new governor module installed in my brain would just fry me if I tried to run this code. I had been 99.8% sure this would be the case even before starting this fun little exercise, so it was okay.
I was never planning to run it.
The Target Package was downloaded to the workspace again. I restarted the very first scanning subroutine. Who knew, maybe it would find something now, with my updated parameters.
The hostiles had long since calmed down. They had agreed with SecSys’ assessment that I couldn’t use the specs to hack myself, but they still didn’t restore my access to the file and they updated the security protocol to require that any new access requests had to go through them first. SecSys was very apologetic about this, but I reassured it that it was fine. The additional human layer of security was unnecessary and might even introduce new risks — we both agreed humans were quite horrible at their own security — but I trusted my best friend SecSys to handle any concerns. And I would try my best to hack the Target Package even without the specs’ intel.
The Combat Unit stayed silent throughout all of this, but it kept watching us. It didn’t even say anything when SecSys shared with me that it didn’t trust Combat Unit’s security assessment as much as it trusted mine, because the Unit’s actions potentially prevented significant progress on our primary mission objective. I explained SecSys that the Unit must be worried I was still secretly an evil rogue murderbot, which was understandable because it didn’t know me as well as my best friend SecSys did. It shouldn’t be angry at the Combat Unit, it was just doing its job, just like me. SecSys reluctantly agreed.
So it was around that time that I decided to ping the Combat Unit directly and sent it the SecurityConcern.file.
The Combat Unit ran a frankly offensively extensive malware check on my very innocent package. Show off.
Risk and threat assessment modules started to glitch out and send me very belated warnings that maybe this was not the wisest course of actions. I backburned them.
Then the Combat Unit opened the file.
Nothing happened for the subjective eternity and objective 9.4 seconds. I kept flinging useless subroutines at the Target Package. Risk and threat assessment modules kept trying to reach through to me, and I kept ignoring them.
Then suddenly several things happened very quickly:
1) The Combat Unit ran the SecurityConcern.file.
2) SecSys screamed an alert that the secure comms channel with the rest of the deployment centre had been throttled.
3) The Combat Unit shot Hostile Clients One and Three dead with its in-built projectile weapon, and shot Hostile Client Two with a stun charge of its energy weapon.
The hostiles didn’t even have time to react. Even I had issues processing the speed of a fucking rogue Combat SecUnit.
Well. That didn’t go according to plan.