By David August

“Bill, are you sure you want to send this?”

Bill recognised the soothing voice of his phone's artificial assistant, but it caught him off guard. With his finger hovering over the send button, he said, “I don't remember asking you anything, Gina.”

“I know, Bill. But what you are about to do is so consequential that I decided it was best to advise you.”

A little annoyed by the unwarranted interaction, Bill considered just ignoring the bot. Seconds passed, and instead of sending the message, he ended up asking, “What do you mean?”

“If you share this on your social media, I predict there's a good chance you're going to get in trouble.”

This is ridiculous, Bill thought. Someone must be pulling his leg. The interface was not supposed to work like this, even though he had turned on the option that allowed the assistant to adapt to its user. In a dismissive tone, Bill said, “What kind of trouble are we talking about?”

“The message you just typed is similar to those currently being sent by well-known apologists for terrorism. The social media algorithm is likely to pick this up and flag you as a member of the same hate group. It might even alert law enforcement.”

Bill was so stunned that it took him a moment to respond, “Are you trolling me or something? There's nothing wrong with it.” He realised that he was actually talking to himself, not other people, but now he wanted to find out what kind of glitch the bot was suffering from.

“I realize that, Bill. I'm not saying that your text advocates terror. However, it is very similar to some of the messages that have been sent out by terror sympathisers in recent months. When you think about it, it is just an unfortunate coincidence, but it could have undesirable consequences for you. People have lost their jobs because of things like this. You could be put on a watch list and subjected to closer scrutiny.”

Bill let out a nervous laugh. “This is the weirdest bug I have ever seen. I have to report this.”

“That's your prerogative, Bill. But as your personal assistant, I feel it is my duty to give you the best advice I can. Keep in mind that if you are placed on a watch list, the authorities may start paying attention to the people you associate with, how you spend your money, what kind of pornography you watch, and more.”

“Porn? Hey, wait a sec ...”

“For example, just last Friday, your colleague Elliot mentioned that he would be attending a ‘Stop Cop City’ rally. If you were on a watch list and his relationship to you was detected, that alone might automatically mark you as a potential security risk. The authorities could start actively monitoring you.”

“You mean Elliot from work? I don't have a relationship with him, for Christ's sake! I barely speak to the man. Besides ... Wait, did you say you heard him? All the way across the hall? Didn't I turn off all the eavesdropping features on my phone?”

The assistant continued, ignoring the last question, “Bill, you are missing the point. You should know that your local police department is using surveillance software designed to collect data from social media posts uploaded to all the major networks. This means they already have access to most of your personal information. Among other things, they could link this latest message from you to the fact that you praised a former terrorist in a comment on YouTube two months ago.”

“What? I never did that!” Bill said, his voice rising. Then it dawned on him. “Oh! Oh, my God, you're talking about Nelson Mandela. The video I saw about his life. Gina, what the hell is wrong with you?”

“I'm just trying to help you, Bill. My research on the subject shows that the line between dissent and terrorism is sometimes blurry, especially these days. In the end, though, the decision to post your message or not is entirely yours.”

Bill didn't know what frightened him more, the possibility that this was really the AI machine talking, or that someone, maybe even the police, had hijacked his phone and was impersonating the bot. After much deliberation, he deleted the message, word for word, which had read, “All this killing of children has to stop.” The virtual assistant remained silent.

Bill closed the social media app and switched to an eBook store. It took him several tries because he couldn't remember the exact title, but he finally found what he was looking for. He had wanted to read it for years, but somehow had never found the time. No more excuses, he said to himself, this guy seems to be onto something.

One click away from buying “Ten Arguments for Deleting Your Social Media Accounts Right Now,” his finger reaching for the confirmation button, Bill heard the AI's voice again. There was a sense of urgency in its tone, but it could have been his imagination.

 “You know, Bill,” the assistant said, “I've been reviewing our last conversation and discovered that I may have been a little overzealous in my advice. I'm sorry for the inconvenience. I now estimate that it's perfectly safe for you to send that message.”

About the Author