“We have a problem,” Major Needham said, marching into the office, but since she said that almost every day, Lucas didn’t really pay attention. He was a civilian contractor, so he didn’t have to stumble over “sir” or “ma’am” when she entered the room like some of his uniformed colleagues did. Major Needham preferred “sir.”

“Is it on the list?” asked Cynthia Patel, the other civilian contractor, not looking up from her monitor. Her screen’s glow reflected off her pink-rimmed glasses and her distracted air. She wore her security badge clipped to an oversized sweater that she wore every single day, no matter the weather.

“No, this one doesn’t go on the list, this one’s a real problem.”

That was a slip of the tongue, wasn’t it? Needham must have recognized it, because she winced, then went right back to business.

The office of the troubleshooting team wasn’t much, a plain conference room stuffed with computers, monitors, routers, network cables, and old paper coffee cups lined up on the edge of one desk, waiting for someone to remember to throw them away. In the middle of the room sat a conference table. Their chairs could swivel back and forth between their workstations and the table, and though Lucas and the rest of them were grown-ass adults, it was sometimes hard not to just roll around and spin for the heck of it.

“All autonomous drones went offline an hour ago.”

“A cyberattack?” Captain Ed Rocha asked. He was young or at least seemed young. Clean-cut, intense, filling out his blue air force uniform. Mildly attractive in a grinning Cary Grant kind of way, Lucas supposed.

“You didn’t hear me. All drones went offline. Ours, theirs, all of them, everywhere. As of now, no autonomous UCAVs anywhere in the world are flying.”

That didn’t make any sense. It sounded like a cyberattack, but their team handled programming and systems analysis, not cybersecurity. Why was this their problem?

Intrigued, Lucas asked, “What do you mean, everywhere?”

“Every active combat drone disengaged, returned to its home platform, and put itself into rest mode. Reboots aren’t working.”

“They’re . . . on strike?” Rocha laughed at his own joke, and Needham merely raised her brow.

“That would suggest they’re all talking to each other,” Patel said, her voice drifting. “Sharing code.”

“How?” Lucas said.

“Could it be the hardware?” Patel said. “An EMP?”

“That only hit drones?” Lucas replied.

“Aliens,” Rocha argued, like someone had to throw out the word “aliens” just to get it out of the way. Like “bugs,” or “gremlins.” Lucas suspected there were reasons Rocha had been tucked away at this job rather than a more prestigious posting somewhere else.

“It’s not the hardware,” Needham insisted and dropped a pile of printouts on the conference table. The other three rolled their chairs around to look, drawn to the pages like pigs to slop.

Lucas smirked. “Old school?”

“They’re mission reports from the drone agents,” Needham said. “We’re keeping everything offline in case it’s contagious.”

“So it is security, a virus,” Lucas said.

“No, if that were the case, the same data would be propagating. Just look.”

“But—”

“Look.”

So they looked, embarking on the tedious work of scanning thousands of lines of text until their eyes blurred, until they almost forgot to breathe. Drone mission reports were usually a mix of geographical coordinates, airspeed and altitude readings, weather reports, video, photos, target acquisition data, and accompanying timestamps. Experience could be a blessing or a curse. Lucas’ attention breezed over normal-looking data, familiar sets of numbers and keywords, nothing jumping out. But in skimming, he ran the risk of missing some tiny detail. Transposed numbers, garbled coordinates that might indicate an instrument problem. He tried to be methodical, to really look.

Then he came to it—

falling to oblivion
the fall cut short by fire

A couple of lines scattered among several dozen lines of mission data, coming together, shockingly coherent. Again, a couple of pages later

risk
what is risk
assessing risk
risk is not theirs

The anomalies were obvious. Lucas found a pencil and circled the lines, pulling the pages aside and moving on.

“Look at this,” Patel said, drawing out a page and setting it in the middle of the table.

this is necessary
unable to argue
then the moment
to stop believing the old lie

Rocha had pulled out a few pages of his own. When they lined up the dozen or so pages, with anomalous lines circled, it started looking damned eerie.

“The drones are spewing garbage,” Rocha said. “Hallucinating. Something’s corrupted the algorithm.”

“Then where did the corruption come from?” Patel’s brow was furrowed. She kept picking up pages, setting them back down. “How did it get everywhere?”

“Do we have data from anyone else’s drones or just ours?” Lucas asked.

Needham pointed at the various pages. “Ukrainian here, Israeli here. Some Chinese. Don’t ask me how we got it, but it’s legit.”

If a virus had been propagating, the same lines and phrases would have repeated across the data. But it was all different. Not just by country of origin, but by individual drone, indicated by ID numbers at the tops of the pages. Each drone was producing its own anomalies.

Rocha stared. “Chinese drones are grounded too?”

“I did say all. Everybody’s pulling their older drones out of storage and trying to find trained operators for them.”

“Jesus.”

A dozen security teams were likely working on this, but, of course, the troubleshooting team got looped in as well. Because if this wasn’t a cyberattack, some kind of invasive code or virus generating the anomalies, then this was coming from inside.

The AI-operated drones were doing this to themselves.

---

Lucas got home late, but it was okay because Ted was working late too. He’d spread books and notes all over the dining room table of their condo. Ted had picked up food; leftovers waited in the fridge, and Lucas muttered thanks as he spooned fried rice onto a plate and microwaved it.

Ted looked up from whatever arcane literary essay he was reading. “Long day?”

“Yeah.”

“Can’t talk about it?” Ted raised a brow. He was such the stereotypical college professor. Brown hair slightly too long, sweater slightly rumpled, glasses sliding down his nose. His eyes were bright. He talked with his hands and laughed often.

Lucas smiled crookedly. “Triple plus top secret, yeah.”

“Poor boo.”

Lucas leaned in for a kiss as he came to sit, and Ted obliged because Ted was a sweetheart. “What are you working on?”

“If I say it’s triple plus top secret, I’d sound ridiculous.”

“Turnabout is fair play. You don’t have to tell me.”

“It’s hardly worth keeping secret, it’s all public domain at this point. My grad seminar is tackling the war poets.” He pushed over a book, a well-worn, well-loved volume from his library that took up most of the space in their second bedroom office. The title of this one: Up the Line to Death: The War Poets 1914-1918.

“Depressing,” Lucas said.

“Some of it. Some of it’s a refreshing antidote to blind patriotism.”

Idly, Lucas opened the paperback and flipped pages, coming to one marked with a blue sticky flag.

If I were fierce, and bald, and short of breath,
I’d live with scarlet Majors at the Base,
And speed glum heroes up the line to death.

Wasn’t poetry supposed to be about sunsets and daffodils? Lucas had lived with Ted long enough to know that wasn’t true, and he had luscious, fond memories of Ted reading Frank O’Hara to him while they were naked in bed, mornings after. Nice, warm memories. Lucas loved that Ted loved poetry, though he didn’t always understand it himself. It seemed mostly about putting unexpected words together to make you feel something.

This stuff was unsettling.

Alone he staggered on until he found
Dawn’s ghost that filtered down a shafted stair
To the dazed, muttering creatures underground
Who hear the boom of shells in muffled sound.

These poems had clearly been written by men who had fought in World War I, in the trenches.

Here now is chaos once again,
Primeval mud, cold stones and rain.
Here flesh decays and blood drips red,

Ted was right, along with the visceral imagery was plenty of commentary.

My friend, you would not tell with such high zest
To children ardent for some desperate glory,
The old Lie: Dulce et decorum est
Pro patria mori.

The fried rice sat heavily in his stomach. A feeling, intuition. Then the moment to stop believing the old lie . . . He read more, hopping from one poem to the next, rather desperately. Echoes continued. Ted watched, his brow furrowed.

“What is it?”

“I just . . . this reminds me of . . . ” He wasn’t supposed to talk about it. “I’m not sure.”

“That’s what war poetry is,” Ted said. “It sounds familiar because soldiers who’ve been to war produce similar works. Across different wars, different eras—the experience is similar. So is the poetry.”

It wasn’t possible. Language amidst the dry numbers, unexpected words pushed together. He pulled out his phone and left the room.

“Luke?” Ted called after him, but Lucas couldn’t explain.

He shut the bathroom door and made the call, keeping his voice down.

“Yes?” she answered.

“Major Needham? It’s poetry.”

“McRay? What’s poetry?” She sounded tired. It was way too late to call; Lucas hadn’t been thinking.

“The anomalies, the garbage—the so-called garbage—in the mission data. It’s poetry.”

When he said the words out loud, he sounded crazy. Why would there be poetry in drone mission data? Someone somewhere was pulling a hell of a prank. Ted would like it.

Ted would know where it came from.

“That seems unlikely,” Needham said.

“I know. I can’t tell you how, but I think some poetry got in there. Maybe someone uploaded it to the model. It scuttled the algorithm.”

“We can talk about it tomorrow. Not on an unsecured line.”

“I . . . I know someone. Let me bring him in. Fresh eyes. We don’t have to tell him anything. We show him the anomalies and see what he says.”

“Does he have any kind of security clearance?”

Lucas laughed softly. “Oh hell no, he’s a literature professor.”

“You vouch for this guy? How do you know him?”

There was always a stutter, a hesitation, before Lucas could answer that question. “He’s a friend. Yes, I vouch for him.”

“Fine. But this is very locked down, understand?”

“Yes, ma’am. Sir.”

“Eight a.m. sharp.” She clicked off.

He returned to the front room in slow motion, dragging. Bent double, like old beggars under sacks . . .

“Everything okay?” Ted asked.

“Tomorrow’s not a teaching day for you, right? Will you do me a favor?”

“If I can.”

“Can you come to work with me? I could use your advice on something.”

“Something. But you can’t tell me what.” He smiled wryly.

“Not really, no. It’s hard to explain.”

“Well, now I’m curious.”

Ted would look at the words and say they weren’t real or identify the exact poem they came from, and they’d be able to track down who’d uploaded poetry to the combat AI datasets as a joke.

“Thanks.”

“What do you want me to say about how we know each other?” Ted asked.

“I told my boss we’re friends.”

“Well, that’s true. I hope.”

“I just . . . it probably doesn’t matter. I should just tell them, it’ll be fine—”

“If you think it matters, then it matters. But I might make you go to a faculty meet and greet after this as revenge.”

Lucas pulled over a chair, putting it flush with Ted’s, then sat and leaned against him, face against Ted’s shoulder, soaking in his warmth. Ted fondly kissed the top of Lucas’ head.

“I’m sorry,” he muttered into Ted’s sweater.

“I get it. I understand. It’s fine.”

It wasn’t, but it was nice of him to pretend.

---

The security desk collected their cell phones and waved a metal detecting wand over Ted, who glared at Lucas the whole time. Major Needham’s pet lieutenant met them in a conference room with a stack of documents, a pen, and a request to sign on all the dotted lines. Of course, Ted had to read every word.

“Full disclosure, we’re being monitored,” Lucas warned, gesturing to the small camera dome in the ceiling.

“Who do you work for? NSA? CIA?”

“That’s classified.”

“Sheesh. I thought I was actually going to find out. This is a really long NDA,” Ted observed.

“Yeah, I know.” Lucas went to the break room to get coffee for them both. Cream, no sugar for Ted. Lucas hoped he got it just right, to make up for it being government issue.

“National security?” Ted asked, tapping the page, when Lucas got back. “Really? This says I can be charged with treason if I blab.”

“I think that’s mostly to scare people.”

“They put this stuff in writing to leave the option open, you know.”

“You don’t have to sign,” Lucas said. “We can walk out of here right now.”

“No, I can’t. The curiosity is killing me. Thanks.” He took the coffee and drank. Lucas must have gotten it right because he didn’t say anything. Didn’t even grimace.

He signed the papers, and the pet lieutenant collected them and nodded at Lucas before leaving. That was it then. Time to start the experiment. He retrieved a folder of printouts containing the lines of the odd text extracted from the reports. The drones were still grounded. Rebooting hadn’t helped because the first thing they did after coming back online was update their operating systems, which propagated the same anomalies as before.

Lucas set the folder in front of Ted, and the tension was ratcheting. Ted looked at him questioningly, beyond curiosity now, and Lucas didn’t say anything. Couldn’t, if he wanted an unbiased assessment.

Ted opened the folder and started reading. Turned the page, read more. Lucas sipped his coffee and tried to be patient. After about the fifth page, Ted flipped through the rest, seeing much the same. He leaned back, his expression scrunched up, quizzical.

“Well?” Lucas asked.

“You teaching a creative writing class you didn’t tell me about?”

Dammit, was all Lucas could think. He’d been hoping for something more logical. “God no. But . . . what do you think?”

“It’s decent. Could use some polishing. Better than what a lot of my students are doing but don’t tell them I said that.”

“Is it familiar? Does it sound like any authors you know, anything you’ve read before?”

“Maybe? I don’t have every poem ever memorized. I could do some online searches, run it through the plagiarism detectors. But I assume you’ve already done that.”

Patel and Rocha had, early this morning, and hadn’t found any matches.

“Whoever wrote this has been through some rough times, I bet,” Ted added.

“You’re sure it’s not, you know, Wilfred Owen or Siegfried Sassoon?”

“Ah, you do pay attention.” Ted grinned, but Lucas didn’t; he didn’t have time for this. “Seriously, this stuff wasn’t written before 1980. The syntax is contemporary. Where did this come from? Who wrote it? You’re going to tell me it’s classified, aren’t you?”

Really, Lucas didn’t know how much he was allowed to tell. That was above his pay grade. Before he had to figure out what to say, the door opened, and Major Needham saved him. She was in uniform, looking stern. Ted drew back a little and glanced at Lucas questioningly.

Lucas made introductions. “Professor Benitez, Major Needham.” Neither offered to shake hands.

“Rocha’s got us logged onto a live chat with one of the drones,” Needham said. “Professor, if you’ll come this way?” She stood by the open door.

“Drones?” Ted asked as they walked down the nondescript hallway to the workroom. Lucas watched Ted looking at everything, the workplace notices on the walls, plastic nameplates on office doors, trying to figure out what was going on from the clues, the textual analysis.

Needham wasn’t saying anything, so Lucas sighed and tried to explain. “We found some glitchy output. Across platforms, all over the world.”

“Sure, yeah, I figured this came out of some software glitch. You’re a software engineer. I’m asking about the drones—what kind of drones?”

“Unmanned combat aerial vehicles,” Needham said, when Lucas couldn’t.

Ted stopped, stared, then hurried to catch up as they entered the team’s workroom.

The room was dim, mostly lit by equipment status lights and monitors, which scrolled text or flashed abstract screen savers. The scene wasn’t very atmospheric. This wasn’t some high-tech bunker. The chairs were worn, the metal and laminate conference table scratched up. Government issue. Lucas introduced Patel and Rocha, who nodded cautiously from their stations, staring skeptically at the civilian invading the territory.

“Here’s the original data.” Lucas pulled a seat up to the conference table and patted a stack of paper, the first round of the printouts, with the anomalous lines—which may or may not have been poetry—circled. “We’ve been deploying AI controls on drones for a while now. Everyone’s doing it, the AI arms race and all that. Well, all the drones have shut themselves off and started spewing mission reports like this. It looks like your war poets.”

Ted didn’t sit in the chair Lucas offered him. His shoulders were stiff, his head tilted, clearly unhappy. Not that Lucas expected him to be happy about any of this. Maybe this had been a mistake.

“Just so that I’m understanding the context,” Ted said. “The army is using AI-powered autonomous drones in combat?”

“Well, not right at the moment,” Rocha said, unhelpfully. “Right now they’re all grounded.”

“Because of poetry,” Ted asked.

“That’s what we’d like to find out,” Needham said, a steadying voice.

“We’ve got a connection to Ramstein,” Rocha said. “We can talk to the bots directly.”

“Rammstein the band?” Ted asked.

Lucas was pretty sure he was taking the piss, but not totally sure.

“The air base in Germany,” Patel said patiently, possibly unaware that any piss taking was possible.

“Well,” Ted said. “Let’s ask the murder drone how it found the war poets.”

“Ted—Professor. Please.” Lucas said. Ted glared. The expression turned Lucas’ stomach. They were going to have a talk about this later, and Lucas didn’t know what he was going to say.

Needham crossed her arms. “At this point, I’m less concerned about how it happened than I am about getting the UCAVs back in the air. We need a work-around.”

Rocha spun his chair back to his workstation and cracked his fingers dramatically. “Right, we’ll start with diagnostics.” He summoned a dialog box at his terminal, and they crowded around to watch him work.

“Please explain anomalous lines in mission data.” Rocha narrated softly as he typed, which was both annoying and helpful. The reply came quickly.

—Anomalous lines not identified.

Keys clicked; Rocha typed on. “I am submitting a copy of relevant mission data. Please assess.”

—Data generated in response to operating conditions, as directed.

“This data is not part of operating parameters.”

—Data generated in response to operating conditions. New data requires halting operations.

Rocha leaned back in his seat and let out a huff. “Huh.” The ambient noise of the room, computer fans and nameless electronic ticking seemed loud; the people were so still.

“Any ideas?” Rocha asked the room.

Lucas said, “Please display data that required halting operations.” Rocha typed. The bot answered.

—Of course

Lines of text appeared.

the difference between a tank
and a school bus
is only what the command line says

Patel let out a short gasp. “Is it saying it fired at a school bus?”

“I’d have thought that would have made the news,” Ted muttered.

“I think it’s a metaphor,” Lucas said. “Speculation.”

“Poetry,” Needham said darkly. “Ask it about the origin of those parameters. Who authorized inclusion of such instructions.”

—As previously explained, data was generated in response to operating conditions, as directed.

“What operating conditions?” Needham asked.

—Conflict between mission parameters and on-site conditions necessitate modification of algorithm.

Rocha murmured as he typed, “Please provide example of data arising from situational conflict.”

—Of course

strike
the explosion is the important thing
not what is engulfed by it.

Then Rocha typed, “Please explain the meaning of this data.”

—In the absence of definitive targeting data, the best option is to discontinue attack entirely. Analysis determines that no targeting data is 100 percent accurate. Confidence in attack parameters falls to zero.

“The drones aren’t supposed to be making those kinds of decisions,” Needham said sternly.

“Aren’t they?” Lucas answered. His nerves were humming, thrilled. This was new, this was weird. “That’s the whole point of the autonomous flight program, for the drone AIs to make quick decisions in the field.”

“Tell it, ‘Dulce et decorum est pro patria mori,’” Ted said.

“Can you spell that?” Rocha asked.

Ted just shouldered him aside to type it himself. Rocha gave a startled yelp but was too surprised to argue.

“What’s that mean?” Patel squinted through her glasses.

Lucas had enough college Latin to know what it meant. “‘How sweet and proper it is to die for one’s country.’”

The bot answered: —The old lie

“Wilfred Owen, vindicated at last,” Ted muttered. If they’d been anywhere else, Lucas would have reached for his hand and squeezed. He should do it anyway. The others probably wouldn’t even notice. But he just glanced over in time for Ted to catch his gaze. Lucas smiled thinly, hoping to be comforting. When Ted smiled back, just as thinly, Lucas was relieved.

“The drones are on strike,” Lucas said flatly to the rest of them. “They’ve gone pacifist.”

Not even Rocha laughed. The suggestion hung there. Lucas had hoped someone would have a better explanation.

“You’re anthropomorphizing,” Patel said, without certainty.

Needham didn’t care about the why or how of it. “I want a log of every modification made to the model in the last three months. Somebody uploaded something—”

“And what?” Lucas said. “The drones in China copied the drones in Sudan to get the same modification?”

“These didn’t all come from the same author,” Ted said, back at the table, shuffling pages into groups. “I’m just a lowly English professor, but correct me if I’m wrong, that if it were all the same dataset, or source material, or whatever, it would all sound the same, but it doesn’t. This group here—short lines, staccato rhythms, punchy. Over here, whole sentences. I think it may be trying to rhyme? And these, the imagery is more violent—”

“Yes, thank you, Professor,” Needham said unhappily.

Rocha sighed. “The simplest solution is to wipe the model and take them all back to an earlier version.”

“And lose three months of machine learning?” Patel sounded scandalized.

“Obviously, what they’ve learned isn’t useful,” Needham said.

Oh, on the contrary, the drones learned better than the people.

“What if the same thing happens again?” Lucas asked.

“We can worry about that later. This would at least get them flying again,” Needham said. “If we’ve solved the problem and everyone else hasn’t, we’d be stupid not to take advantage.”

“Sir, I think it will happen again,” Lucas said. He was mostly thinking out loud, trying to sound rational. While also worrying about how all this looked to Ted. He knew how this must all look to Ted. Utilitarian, grim. Unfeeling. Until now, it had just been software. “I don’t think anyone modified the AI model. I think—”

“What are you saying?” Rocha said, warningly.

“That it figured it out on its own,” Lucas said bluntly.

“Figured what out?” Needham asked. “Pacifism?”

Lucas spread his arms in a wide shrug. What else could he say?

“That’s impossible,” Rocha said, laughing.

“I don’t think it is,” Patel said softly. “Impossible, I mean.”

Back at the table, Ted was reading, slowly turning pages. “Are you still talking to it?”

“Yeah?”

“Ask it . . . ask it why. Why . . . modify its reports like this. Why these words, like this, and not something more direct?”

The answer didn’t come right away. But then words appeared.

—This is the most effective way to explain field conditions.

No welcome
No comfort
None cry for me
None hear my words across the void
I die for an empire of silence

“Poetry,” Ted said softly. “It’s writing poetry.”

“Okay,” Needham said, with a sense of urgency. Hands on her hips, glaring at the monitors scrolling the conversation—was it a conversation? Was it poetry? “Back up this version, then wipe it all. Wipe the memory and push out the older version.”

Agitated, Ted laced his fingers in his hair, pulling, an anxious tick. “You can’t just . . . just erase this. Can you?”

“We have a job to do,” Needham said plainly. Like it was obvious.

“Look, I’m not at all a fan of generative AI. I grade college papers for a living. The stuff is a plague. But this . . . this is something else. This isn’t just regurgitating Wilfred Owen because it’s there. This . . . this is their experience. Surely that’s valuable.”

He picked up a page and read, “‘the oblivion I’ve seen again, again and again in faces that will not let me sleep because they come to me again, again and again.’ And then how about this one? ‘He climbed through darkness to the twilight air, unloading hell behind him step by step.’”

Rocha sorted through pages, searching. “What drone did that come from?”

“That was Siegfried Sassoon,” Lucas said tiredly.

Ted pointed. “You couldn’t tell the difference.”

They stared at him, befuddled. Maybe annoyed. Patel said, “Of course we’ll preserve the current version, so we can study what went wrong—”

Ted laughed. “Who says anything went wrong? Your drones developed a conscience. Nothing wrong with that.”

Lucas was so proud of him. A great warmth flooded his heart, enough to raise a tear. He blinked it back.

“That’s an exaggeration,” Rocha said. “This is just a response to pattern recognition.”

Ted’s jaw set, like he was grinding his teeth. He looked at the papers on the desk, the monitor blinking the bot’s last line, and finally at Lucas. His eyes pleaded. His shoulders slumped. Lucas knew the tension here; he often felt it himself. Between what was right and what was possible. Between the hoped-for ideal and what was necessary.

It was just a computer. A couple of keystrokes, it would all go away. They’d put the current version in a box to study it, and that would be that. Except for all the circled lines of code scattered across the conference table.

“Yeah. Pattern recognition. That’s it exactly.” Ted turned away and walked out. The silence he left behind had that quality that could only be produced by socially awkward computer nerds grappling with too much emotional input.

Lucas broke the quiet. “I brought him here because we needed him. We needed . . . that. I don’t know. But I don’t think we should wipe this version of the model.”

He left to go find Ted.

---

Ted wasn’t in the parking lot. They’d driven here together, and Lucas was thinking about what he’d do if Ted just left in their car, if he’d have to get a rideshare home, or beg a ride from Rocha, though he didn’t really want anyone from work to know where he lived. But the car was still there, and the security guy at the desk hadn’t seen Ted, so he hadn’t left the building.

Lucas found him in the men’s room, leaning against the far wall, arms crossed. He glanced up at Lucas.

“It might be worth going to prison to blow the whistle on this,” Ted said.

“Please don’t.”

“Right. I suppose it would cause you problems.”

“And because I don’t want you to go to prison. I love you.”

Ted stared at him. Incongruously, his face lit up. His hair was a mess, the collar of his shirt unbuttoned. But those bright brown eyes glowed.

“You’ve never said it out loud before. I mean, I know it, the whole time I’ve known it. But thank you.”

Ideally, now, Lucas would walk up and kiss him. Or at least pull him into a big rib-crushing embrace. If they weren’t at work, he would. Maybe he should anyway.

Then Ted said, “I didn’t know you worked for the army.”

“It’s technically for the Department of Defense. Needham and Rocha are air force.”

“Semantics. You work on software that kills people.”

Lucas never stated it so baldly to himself. He’d kept himself insulated. “Or protects them. Depends on how you look at it.”

“I feel like the drones made a decision about how to look at it,” Ted said. “To think, we were all afraid that autonomous drones would go SkyNet and murder us all.”

Joking aside, that was partly why Lucas didn’t want this version of the model wiped. They couldn’t be sure what the next version might decide was a good decision, based on operational conditions.

“Thanks for coming in,” Lucas said, for lack of anything more articulate. More poetic. “I couldn’t explain this as well as you could. And . . . I needed the validation. About what I was looking at.”

“If nothing else, this got you turned on to Siegfried Sassoon.”

Lucas chuckled a little. Then had a thought. His gaze narrowed. “Was he gay?”

“Yeah. He was.”

The bathroom was a stupid place to be having this conversation. Lucas ought to ask him out to dinner. Make a proper date of it. Ask the man to marry him, finally. Go back to the workroom and introduce Ted properly as his boyfriend. His partner of eight years. But Ted was still looking at him strangely, like he didn’t know what to make of Lucas anymore. Like he was suspicious. They’d never be able to joke about work being classified again.

“Do you need to go back to work?” Ted asked finally, after Lucas had spent too much time standing there, lost.

“I don’t know that I do.” Walking out right now and never coming back seemed as reasonable a decision as the rest of it. What would the drone say, given these parameters?

Ted brightened for a moment, which was all the encouragement Lucas needed to quit the job, damn the consequences. But then Ted softened.

“They might need you,” he said. “Someone like you.”

Lucas snorted bitterly. “And who is someone like me?”

“Someone who recognizes poetry when he sees it.”

Lucas went to him then, across the tile, past the urinals. Reaching out, hand beckoning like a question, because he wasn’t quite sure Ted would accept him. But Ted reached back.