Recently wrote arguing that AI should have to identify itself as such. Would love the communities feedback!

# Who Goes There? The Case for Compelling AI Identification Source: [https://socialmediaatlaw.substack.com/p/who-goes-there-the-case-for-compelling?r=f04xx](https://socialmediaatlaw.substack.com/p/who-goes-there-the-case-for-compelling?r=f04xx) It was not that long ago that if you told someone that a machine would soon be capable of writing articles, answering legal questions, creating realistic images, carrying on lengthy conversations, and flooding social media with human\-like comments, they probably would have assumed you had been watching too much science fiction\. Today, that future is here and with it has come seemingly endless applications\. Every day normal people, criminals, and companies alike are finding new ways to apply this technology\. Yet despite all of this attention being paid to Artificial Intelligence, we are largely ignoring what may be the most important question of all: Should people have a right to know when they are interacting with a machine? Colorado recently answered that question with a qualified “yes\.” [Beginning January 1st, 2027, Colorado will require employers using certain forms of Artificial Intelligence in hiring and compensation decisions to notify applicants, explain adverse decisions, and, in some circumstances, offer human review](https://leg.colorado.gov/bills/sb26-189)\. Whether one supports the law or opposes it, Colorado lawmakers should at least get credit for recognizing an important principle: when Artificial Intelligence materially influences an important decision in a person’s life, people have a legitimate interest in knowing it\. That principle should not stop at the workplace\. Should we as a society not have a endowed right to know if artificial intelligence is being used to persuade, influence, or otherwise control us? Why shouldn’t a person not know whether the customer service representative they are chatting with is actually a chatbot? Why shouldn’t we know whether the glowing restaurant review we are reading was written by a person or generated by software? Why shouldn’t voters know whether the social media account arguing with me about politics is a in fact a person with thoughts, opinions, and experiences of their own or a machine designed to upset, engage, or otherwise manipulate them? Artificial intelligence is here\. Bans, government censorship regimes, and content moderation mandates can be debated and discussed but one thing that society should agree on in this instances, at a minimum, is transparency\. In recent years, many technology debates have become arguments about censorship\. What should platforms remove? What should they leave up? Who gets to decide what is misinformation? Which experts should be trusted? Who even is an expert? Reasonable people can disagree on those questions\. The proposal here is much narrower and in no way radical\. In fact, it is built upon one of the most common principles of American law: the right to information\. We require nutritional labels on food, disclosures in securities offerings, identification of sponsors in many political advertisements, and disclosure of conflicts of interest in countless professional settings to name but a few\. Why not require software that may be informing people about the content of that food, the stabilizability of those securities, the positions of those politicians, and the requirements of those professions to say it is software so that a user can better gauge its credibility? As Artificial Intelligence becomes increasingly capable of imitating human beings, whether a message was generated by a person or a machine is rapidly becoming one of those facts\. Imagine a political candidate whose campaign deploys ten thousand AI\-generated accounts to praise the candidate, attack opponents, and create the appearance of widespread grassroots support\. Imagine a corporation using AI\-generated reviews to flood the internet with positive feedback about its products\. Imagine a foreign government operating thousands of AI\-powered accounts designed to influence public opinion on controversial issues\. Worst still, imagine a scammer using an AI powered voice model to tell the little old lady down the street exactly what she needs to hear to hand over her social security number\. None of these scenarios require Artificial Intelligence to become sentient, none require science fiction\. We need not be fighting Terminators or running from Westworld style hosts\. All require only the technology that already exists today\. Predictably, some will object that people should simply learn to recognize AI\-generated content making the usually persuasive argument that “complexity is a subsidy”, that new laws would hinder the growth of the AI industry\. This is undeniably true but like any regulation political leaders must weigh the harm the unbridled market could cause against the benefits of an intelligently imposed structure\. Disclosure laws have never depended on consumers being incapable of discovering the truth but rather on ensuring the public have the necessary information to make the best choices, to provide the metaphorical clay with which to make bricks to build our lives\. Investors can perform their own due diligence\. We still require disclosures\. Consumers can research food ingredients\. We still require labels\. Voters can investigate campaign funding\. We still require reporting requirements\. Criminals by their nature will flout the law in order to commit fraud but if the underlying software has rules built in forcing it to identify itself when asked the intent of the scammer will, at a minimum, be frustrated if not stopped\. The purpose of disclosure is not to eliminate personal responsibility\. The purpose is to ensure that people have access to material information before making decisions\. Further, transparency requirements rarely destroy industries\. They typically strengthen public confidence in them\. Seatbelts did not kill the automobile\. Nutritional labels did not kill the food industry\. Financial disclosures did not kill capital markets\. Requiring Artificial Intelligence to identify itself would not stop innovation\. It would simply ensure that innovation develops in an environment of transparency and foster greater faith in these systems and the companies that create or use them\. What makes this proposal particularly appealing is that it offers something for nearly every constituency which is particularly hard to do in the current climate: Conservatives, wary of government censorship, can support disclosure because it regulates transparency rather than speech itself\. The government would not be deciding what Artificial Intelligence may say, merely require honesty about the speaker\. Liberals concerned about misinformation and manipulation can support disclosure because it increases transparency without empowering regulators to determine truth or falsity\. The focus remains on the source, not the content\. Libertarians, perhaps the group least likely to support any regulations, agree that informed consent and informed decision\-making are foundational principles of a free society\. People remain free to engage with Artificial Intelligence\. They are simply given the information necessary to make that choice intelligently\. Businesses too should support disclosure because it provides a clear and predictable framework that is far less burdensome than many of the alternative regulatory proposals currently being discussed\. Companies remain free to use AI\. They simply disclose that they are doing so\. Placing a minor control now may prevent the need for more onerous regulations later\. In a political environment where almost every technology debate devolves into a fight over censorship, that kind of consensus is increasingly rare\. Colorado’s hiring law may have been aimed at employment decisions, but the principle underlying it is much broader\. People care whether Artificial Intelligence is influencing important aspects of their lives\. They care whether they are being evaluated by a machine, they care whether they are being advised by a machine, and increasingly, they care whether they are being persuaded by a machine\. For decades, internet users have been taught \(in theory if not in practice\) to ask whether information is true\. The age of Artificial Intelligence adds a new threshold question: is the information being provided even coming from a human being? Before we decide whether to trust a message, we may first need to know whether anyone actually wrote it\. In a free society, that seems like something people have a right to know\. [](https://substackcdn.com/image/fetch/$s_!JVAW!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb524dd93-8d6b-423d-9125-4d6d2d761350_1024x1024.jpeg)