## November 15, 2023

by Bruce Schneier
Fellow and Lecturer, Harvard Kennedy School
schneier@schneier.com
https://www.schneier.com

A free monthly newsletter providing summaries, analyses, insights, and commentaries on security: computer and otherwise.

For back issues, or to subscribe, visit Crypto-Gram’s web page.

These same essays and news items appear in the Schneier on Security blog, along with a lively and intelligent comment section. An RSS feed is available.

### Coin Flips Are Biased

[2023.10.16] Experimental result:

Many people have flipped coins but few have stopped to ponder the statistical and physical intricacies of the process. In a preregistered study we collected 350,757 coin flips to test the counterintuitive prediction from a physics model of human coin tossing developed by Persi Diaconis. The model asserts that when people flip an ordinary coin, it tends to land on the same side it started—Diaconis estimated the probability of a same-side outcome to be about 51%.

And the final paragraph:

Could future coin tossers use the same-side bias to their advantage? The magnitude of the observed bias can be illustrated using a betting scenario. If you bet a dollar on the outcome of a coin toss (i.e., paying 1 dollar to enter, and winning either 0 or 2 dollars depending on the outcome) and repeat the bet 1,000 times, knowing the starting position of the coin toss would earn you 19 dollars on average. This is more than the casino advantage for 6 deck blackjack against an optimal-strategy player, where the casino would make 5 dollars on a comparable bet, but less than the casino advantage for single-zero roulette, where the casino would make 27 dollars on average. These considerations lead us to suggest that when coin flips are used for high-stakes decision-making, the starting position of the coin is best concealed.

Boing Boing post.

### Security Vulnerability of Switzerland’s E-Voting System

[2023.10.17] Online voting is insecure, period. This doesn’t stop organizations and governments from using it. (And for low-stakes elections, it’s probably fine.) Switzerland—not low stakes—uses online voting for national elections. Andrew Appel explains why it’s a bad idea:

Last year, I published a 5-part series about Switzerland’s e-voting system. Like any internet voting system, it has inherent security vulnerabilities: if there are malicious insiders, they can corrupt the vote count; and if thousands of voters’ computers are hacked by malware, the malware can change votes as they are transmitted. Switzerland “solves” the problem of malicious insiders in their printing office by officially declaring that they won’t consider that threat model in their cybersecurity assessment.

But it also has an interesting new vulnerability:

The Swiss Post e-voting system aims to protect your vote against vote manipulation and interference. The goal is to achieve this even if your own computer is infected by undetected malware that manipulates a user vote. This protection is implemented by special return codes (Prüfcode), printed on the sheet of paper you receive by physical mail. Your computer doesn’t know these codes, so even if it’s infected by malware, it can’t successfully cheat you as long as, you follow the protocol.

Unfortunately, the protocol isn’t explained to you on the piece of paper you get by mail. It’s only explained to you online, when you visit the e-voting website. And of course, that’s part of the problem! If your computer is infected by malware, then it can already present to you a bogus website that instructs you to follow a different protocol, one that is cheatable. To demonstrate this, I built a proof-of-concept demonstration.

Appel again:

Kuster’s fake protocol is not exactly what I imagined; it’s better. He explains it all in his blog post. Basically, in his malware-manipulated website, instead of displaying the verification codes for the voter to compare with what’s on the paper, the website asks the voter to enter the verification codes into a web form. Since the website doesn’t know what’s on the paper, that web-form entry is just for show. Of course, Kuster did not employ a botnet virus to distribute his malware to real voters! He keeps it contained on his own system and demonstrates it in a video.

Again, the solution is paper. (Here I am saying that in 2004.) And, no, blockchain does not help—it makes security worse.

### Analysis of Intellexa’s Predator Spyware

[2023.10.18] Amnesty International has published a comprehensive analysis of the Predator government spyware products.

These technologies used to be the exclusive purview of organizations like the NSA. Now they’re available to every country on the planet—democratic, nondemocratic, authoritarian, whatever—for a price. This is the legacy of not securing the Internet when we could have.

### Former Uber CISO Appealing His Conviction

[2023.10.19] Joe Sullivan, Uber’s CISO during their 2016 data breach, is appealing his conviction.

Prosecutors charged Sullivan, whom Uber hired as CISO after the 2014 breach, of withholding information about the 2016 incident from the FTC even as its investigators were scrutinizing the company’s data security and privacy practices. The government argued that Sullivan should have informed the FTC of the 2016 incident, but instead went out of his way to conceal it from them.

Prosecutors also accused Sullivan of attempting to conceal the breach itself by paying \$100,000 to buy the silence of the two hackers behind the compromise. Sullivan had characterized the payment as a bug bounty similar to ones that other companies routinely make to researchers who report vulnerabilities and other security issues to them. His lawyers pointed out that Sullivan had made the payment with the full knowledge and blessing of Travis Kalanick, Uber’s CEO at the time, and other members of the ride-sharing giant’s legal team.

But prosecutors described the payment and an associated nondisclosure agreement that Sullivan’s team wanted the hackers to sign as an attempt to cover up what was in effect a felony breach of Uber’s network.

[…]

Sullivan’s fate struck a nerve with many peers and others in the industry who perceived CISOs as becoming scapegoats for broader security failures at their companies. Many argued and continue to argue that Sullivan acted with the full knowledge of his supervisors but in the end became the sole culprit for the breach and the associated failures for which he was charged. They believed that if Sullivan could be held culpable for his failure to report the 2016 breach to the FTC – and for the alleged hush payment—then so should Kalanick at the very least, and probably others as well.

It’s an argument that Sullivan’s lawyers once again raised in their appeal of the obstruction conviction this week. “Despite the fact that Mr. Sullivan was not responsible at Uber for the FTC’s investigation, including the drafting or signing any of the submissions to the FTC, the government singled him out among over 30 of his co-employees who all had information that Mr. Sullivan is alleged to have hidden from the FTC,” Swaminathan said.

I have some sympathy for that view. Sullivan was almost certainly scapegoated here. But I do want executives personally liable for what their company does. I don’t know enough about the details to have an opinion in this particular case.

### AI and US Election Rules

[2023.10.20] If an AI breaks the rules for you, does that count as breaking the rules? This is the essential question being taken up by the Federal Election Commission this month, and public input is needed to curtail the potential for AI to take US campaigns (even more) off the rails.

At issue is whether candidates using AI to create deepfaked media for political advertisements should be considered fraud or legitimate electioneering. That is, is it allowable to use AI image generators to create photorealistic images depicting Trump hugging Anthony Fauci? And is it allowable to use dystopic images generated by AI in political attack ads?

For now, the answer to these questions is probably “yes.” These are fairly innocuous uses of AI, not any different than the old-school approach of hiring actors and staging a photoshoot, or using video editing software. Even in cases where AI tools will be put to scurrilous purposes, that’s probably legal in the US system. Political ads are, after all, a medium in which you are explicitly permitted to lie.

The concern over AI is a distraction, but one that can help draw focus to the real issue. What matters isn’t how political content is generated; what matters is the content itself and how it is distributed.

Future uses of AI by campaigns go far beyond deepfaked images. Campaigns will also use AI to personalize communications. Whereas the previous generation of social media microtargeting was celebrated for helping campaigns reach a precision of thousands or hundreds of voters, the automation offered by AI will allow campaigns to tailor their advertisements and solicitations to the individual.

Most significantly, AI will allow digital campaigning to evolve from a broadcast medium to an interactive one. AI chatbots representing campaigns are capable of responding to questions instantly and at scale, like a town hall taking place in every voter’s living room, simultaneously. Ron DeSantis’ presidential campaign has reportedly already started using OpenAI’s technology to handle text message replies to voters.

At the same time, it’s not clear whose responsibility it is to keep US political advertisements grounded in reality—if it is anyone’s. The FEC’s role is campaign finance, and is further circumscribed by the Supreme Court’s repeated stripping of its authorities. The Federal Communications Commission has much more expansive responsibility for regulating political advertising in broadcast media, as well as political robocalls and text communications. However, the FCC hasn’t done much in recent years to curtail political spam. The Federal Trade Commission enforces truth in advertising standards, but political campaigns have been largely exempted from these requirements on First Amendment grounds.

To further muddy the waters, much of the online space remains loosely regulated, even as campaigns have fully embraced digital tactics. There are still insufficient disclosure requirements for digital ads. Campaigns pay influencers to post on their behalf to circumvent paid advertising rules. And there are essentially no rules beyond the simple use of disclaimers for videos that campaigns post organically on their own websites and social media accounts, even if they are shared millions of times by others.

Almost everyone has a role to play in improving this situation.

Let’s start with the platforms. Google announced earlier this month that it would require political advertisements on YouTube and the company’s other advertising platforms to disclose when they use AI images, audio, and video that appear in their ads. This is to be applauded, but we cannot rely on voluntary actions by private companies to protect our democracy. Such policies, even when well-meaning, will be inconsistently devised and enforced.

The FEC should use its limited authority to stem this coming tide. The FEC’s present consideration of rulemaking on this issue was prompted by Public Citizen, which petitioned the Commission to “clarify that the law against ‘fraudulent misrepresentation’ (52 U.S.C. §30124) applies to deliberately deceptive AI-produced content in campaign communications.” The FEC’s regulation against fraudulent misrepresentation (C.F.R. §110.16) is very narrow; it simply restricts candidates from pretending to be speaking on behalf of their opponents in a “damaging” way.

Extending this to explicitly cover deepfaked AI materials seems appropriate. We should broaden the standards to robustly regulate the activity of fraudulent misrepresentation, whether the entity performing that activity is AI or human—but this is only the first step. If the FEC takes up rulemaking on this issue, it could further clarify what constitutes “damage.” Is it damaging when a PAC promoting Ron DeSantis uses an AI voice synthesizer to generate a convincing facsimile of the voice of his opponent Donald Trump speaking his own Tweeted words? That seems like fair play. What if opponents find a way to manipulate the tone of the speech in a way that misrepresents its meaning? What if they make up words to put in Trump’s mouth? Those use cases seem to go too far, but drawing the boundaries between them will be challenging.

Congress has a role to play as well. Senator Klobuchar and colleagues have been promoting both the existing Honest Ads Act and the proposed REAL Political Ads Act, which would expand the FEC’s disclosure requirements for content posted on the Internet and create a legal requirement for campaigns to disclose when they have used images or video generated by AI in political advertising. While that’s worthwhile, it focuses on the shiny object of AI and misses the opportunity to strengthen law around the underlying issues. The FEC needs more authority to regulate campaign spending on false or misleading media generated by any means and published to any outlet. Meanwhile, the FEC’s own Inspector General continues to warn Congress that the agency is stressed by flat budgets that don’t allow it to keep pace with ballooning campaign spending.

It is intolerable for such a patchwork of commissions to be left to wonder which, if any of them, has jurisdiction to act in the digital space. Congress should legislate to make clear that there are guardrails on political speech and to better draw the boundaries between the FCC, FEC, and FTC’s roles in governing political speech. While the Supreme Court cannot be relied upon to uphold common sense regulations on campaigning, there are strategies for strengthening regulation under the First Amendment. And Congress should allocate more funding for enforcement.

The FEC has asked Congress to expand its jurisdiction, but no action is forthcoming. The present Senate Republican leadership is seen as an ironclad barrier to expanding the Commission’s regulatory authority. Senate Majority Leader Mitch McConnell has a decades-long history of being at the forefront of the movement to deregulate American elections and constrain the FEC. In 2003, he brought the unsuccessful Supreme Court case against the McCain-Feingold campaign finance reform act (the one that failed before the Citizens United case succeeded).

The most impactful regulatory requirement would be to require disclosure of interactive applications of AI for campaigns—and this should fall under the remit of the FCC. If a neighbor texts me and urges me to vote for a candidate, I might find that meaningful. If a bot does it under the instruction of a campaign, I definitely won’t. But I might find a conversation with the bot—knowing it is a bot—useful to learn about the candidate’s platform and positions, as long as I can be confident it is going to give me trustworthy information.

The FCC should enter rulemaking to expand its authority for regulating peer-to-peer (P2P) communications to explicitly encompass interactive AI systems. And Congress should pass enabling legislation to back it up, giving it authority to act not only on the SMS text messaging platform, but also over the wider Internet, where AI chatbots can be accessed over the web and through apps.

And the media has a role. We can still rely on the media to report out what videos, images, and audio recordings are real or fake. Perhaps deepfake technology makes it impossible to verify the truth of what is said in private conversations, but this was always unstable territory.

What is your role? Those who share these concerns could submit a comment to the FEC’s open public comment process before October 16, urging it to use its available authority. We all know government moves slowly, but a show of public interest is necessary to get the wheels moving.

Ultimately, all these policy changes serve the purpose of looking beyond the shiny distraction of AI to create the authority to counter bad behavior by humans. Remember: behind every AI is a human who should be held accountable.

This essay was written with Nathan Sanders, and was previously published on the Ash Center website.

### Child Exploitation and the Crypto Wars

[2023.10.23] Susan Landau published an excellent essay on the current justification for the government breaking end-to-end-encryption: child sexual abuse and exploitation (CSAE). She puts the debate into historical context, discusses the problem of CSAE, and explains why breaking encryption isn’t the solution.

### EPA Won’t Force Water Utilities to Audit Their Cybersecurity

[2023.10.24] The industry pushed back:

Despite the EPA’s willingness to provide training and technical support to help states and public water system organizations implement cybersecurity surveys, the move garnered opposition from both GOP state attorneys and trade groups.

Republican state attorneys that were against the new proposed policies said that the call for new inspections could overwhelm state regulators. The attorney generals of Arkansas, Iowa and Missouri all sued the EPA—claiming the agency had no authority to set these requirements. This led to the EPA’s proposal being temporarily blocked back in June.

So now we have a piece of our critical infrastructure with substandard cybersecurity. This seems like a really bad outcome.

### Microsoft is Soft-Launching Security Copilot

[2023.10.25] Microsoft has announced an early access program for its LLM-based security chatbot assistant: Security Copilot.

I am curious whether this thing is actually useful.

### New NSA Information from (and about) Snowden

MacAskill, who shared the Pulitzer Prize for Public Service with Glenn Greenwald and Laura Poitras for their journalistic work on the Snowden files, retired from The Guardian in 2018. He told Computer Weekly that:

• As far as he knows, a copy of the documents is still locked in the New York Times office. Although the files are in the New York Times office, The Guardian retains responsibility for them.
• As to why the New York Times has not published them in a decade, MacAskill maintains “this is a complicated issue.” “There is, at the very least, a case to be made for keeping them for future generations of historians,” he said.
• Why was only 1% of the Snowden archive published by the journalists who had full access to it? Ewen MacAskill replied: “The main reason for only a small percentage—though, given the mass of documents, 1% is still a lot—was diminishing interest.”

[…]

The Guardian’s journalist did not recall seeing the three revelations published by Computer Weekly, summarized below:

• The NSA listed Cavium, an American semiconductor company marketing Central Processing Units (CPUs)—the main processor in a computer which runs the operating system and applications—as a successful example of a “SIGINT-enabled” CPU supplier. Cavium, now owned by Marvell, said it does not implement back doors for any government.
• The NSA compromised lawful Russian interception infrastructure, SORM. The NSA archive contains slides showing two Russian officers wearing jackets with a slogan written in Cyrillic: “You talk, we listen.” The NSA and/or GCHQ has also compromised key lawful interception systems.
• Among example targets of its mass-surveillance programme, PRISM, the NSA listed the Tibetan government in exile.

Those three pieces of info come from Jake Appelbaum’s PhD thesis.

### Messaging Service Wiretap Discovered through Expired TLS Cert

[2023.10.27] Fascinating story of a covert wiretap that was discovered because of an expired TLS certificate:

The suspected man-in-the-middle attack was identified when the administrator of jabber.ru, the largest Russian XMPP service, received a notification that one of the servers’ certificates had expired.

However, jabber.ru found no expired certificates on the server, as explained in a blog post by ValdikSS, a pseudonymous anti-censorship researcher based in Russia who collaborated on the investigation.

The expired certificate was instead discovered on a single port being used by the service to establish an encrypted Transport Layer Security (TLS) connection with users. Before it had expired, it would have allowed someone to decrypt the traffic being exchanged over the service.

### Hacking Scandinavian Alcohol Tax

[2023.10.30] The islands of Åland are an important tax hack:

Although Åland is part of the Republic of Finland, it has its own autonomous parliament. In areas where Åland has its own legislation, the group of islands essentially operates as an independent nation.

This allows Scandinavians to avoid the notoriously high alcohol taxes:

Åland is a member of the EU and its currency is the euro, but Åland’s relationship with the EU is regulated by way of a special protocol. In order to maintain the important sale of duty-free goods on ferries operating between Finland and Sweden, Åland is not part of the EU’s VAT area.

Basically, ferries between the two countries stop at the island, and people stock up—I mean really stock up, hand trucks piled with boxes—on tax-free alcohol. Åland gets the revenue, and presumably docking fees.

The purpose of the special status of the Åland Islands was to maintain the right to tax free sales in the ship traffic. The ship traffic is of vital importance for the province’s communication, and the intention was to support the economy of the province this way.

### The Future of Drone Warfare

[2023.10.31] Ukraine is using \$400 drones to destroy tanks:

Facing an enemy with superior numbers of troops and armor, the Ukrainian defenders are holding on with the help of tiny drones flown by operators like Firsov that, for a few hundred dollars, can deliver an explosive charge capable of destroying a Russian tank worth more than \$2 million.

[…]

A typical FPV weighs up to one kilogram, has four small engines, a battery, a frame and a camera connected wirelessly to goggles worn by a pilot operating it remotely. It can carry up to 2.5 kilograms of explosives and strike a target at a speed of up to 150 kilometers per hour, explains Pavlo Tsybenko, acting director of the Dronarium military academy outside Kyiv.

“This drone costs up to \$400 and can be made anywhere. We made ours using microchips imported from China and details we bought on AliExpress. We made the carbon frame ourselves. And, yeah, the batteries are from Tesla. One car has like 1,100 batteries that can be used to power these little guys,” Tsybenko told POLITICO on a recent visit, showing the custom-made FPV drones used by the academy to train future drone pilots.

“It is almost impossible to shoot it down,” he said. “Only a net can help. And I predict that soon we will have to put up such nets above our cities, or at least government buildings, all over Europe.”

Science fiction authors have been writing about drone swarms for decades. Now they are reality. Tanks today. Soon it will be ships (probably with more expensive drones). Feels like this will be a major change in warfare.

### Spyware in India

[2023.11.02] Apple has warned leaders of the opposition government in India that their phones are being spied on:

Multiple top leaders of India’s opposition parties and several journalists have received a notification from Apple, saying that “Apple believes you are being targeted by state-sponsored attackers who are trying to remotely compromise the iPhone associated with your Apple ID ….”

AccessNow puts this in context:

For India to uphold fundamental rights, authorities must initiate an immediate independent inquiry, implement a ban on the use of rights-abusing commercial spyware, and make a commitment to reform the country’s surveillance laws. These latest warnings build on repeated instances of cyber intrusion and spyware usage, and highlights the surveillance impunity in India that continues to flourish despite the public outcry triggered by the 2019 Pegasus Project revelations.

### New York Increases Cybersecurity Rules for Financial Companies

[2023.11.03] Another example of a large and influential state doing things the federal government won’t:

Boards of directors, or other senior committees, are charged with overseeing cybersecurity risk management, and must retain an appropriate level of expertise to understand cyber issues, the rules say. Directors must sign off on cybersecurity programs, and ensure that any security program has “sufficient resources” to function.

In a new addition, companies now face significant requirements related to ransom payments. Regulated firms must now report any payment made to hackers within 24 hours of that payment.

### Crashing iPhones with a Flipper Zero

[2023.11.06] The Flipper Zero is an incredibly versatile hacking device. Now it can be used to crash iPhones in its vicinity by sending them a never-ending stream of pop-ups.

These types of hacks have been possible for decades, but they require special equipment and a fair amount of expertise. The capabilities generally required expensive SDRs—short for software-defined radios—that, unlike traditional hardware-defined radios, use firmware and processors to digitally re-create radio signal transmissions and receptions. The \$200 Flipper Zero isn’t an SDR in its own right, but as a software-controlled radio, it can do many of the same things at an affordable price and with a form factor that’s much more convenient than the previous generations of SDRs.

### Spaf on the Morris Worm

[2023.11.07] Gene Spafford wrote an essay reflecting on the Morris Worm of 1988—thirty-five years ago. His lessons from then are still applicable today.

### Decoupling for Security

[2023.11.08] This is an excerpt from a longer paper. You can read the whole thing (complete with sidebars and illustrations) here.

Our message is simple: it is possible to get the best of both worlds. We can and should get the benefits of the cloud while taking security back into our own hands. Here we outline a strategy for doing that.

### What Is Decoupling?

In the last few years, a slew of ideas old and new have converged to reveal a path out of this morass, but they haven’t been widely recognized, combined, or used. These ideas, which we’ll refer to in the aggregate as “decoupling,” allow us to rethink both security and privacy.

Here’s the gist. The less someone knows, the less they can put you and your data at risk. In security this is called Least Privilege. The decoupling principle applies that idea to cloud services by making sure systems know as little as possible while doing their jobs. It states that we gain security and privacy by separating private data that today is unnecessarily concentrated.

To unpack that a bit, consider the three primary modes for working with our data as we use cloud services: data in motion, data at rest, and data in use. We should decouple them all.

Our data is in motion as we exchange traffic with cloud services such as videoconferencing servers, remote file-storage systems, and other content-delivery networks. Our data at rest, while sometimes on individual devices, is usually stored or backed up in the cloud, governed by cloud provider services and policies. And many services use the cloud to do extensive processing on our data, sometimes without our consent or knowledge. Most services involve more than one of these modes.

To ensure that cloud services do not learn more than they should, and that a breach of one does not pose a fundamental threat to our data, we need two types of decoupling. The first is organizational decoupling: dividing private information among organizations such that none knows the totality of what is going on. The second is functional decoupling: splitting information among layers of software. Identifiers used to authenticate users, for example, should be kept separate from identifiers used to connect their devices to the network.

In designing decoupled systems, cloud providers should be considered potential threats, whether due to malice, negligence, or greed. To verify that decoupling has been done right, we can learn from how we think about encryption: you’ve encrypted properly if you’re comfortable sending your message with your adversary’s communications system. Similarly, you’ve decoupled properly if you’re comfortable using cloud services that have been split across a noncolluding group of adversaries.

This essay was written with Barath Raghavan, and previously appeared in IEEE Spectrum.

### Online Retail Hack

[2023.11.09] Selling miniature replicas to unsuspecting shoppers:

Online marketplaces sell tiny pink cowboy hats. They also sell miniature pencil sharpeners, palm-size kitchen utensils, scaled-down books and camping chairs so small they evoke the Stonehenge scene in “This Is Spinal Tap.” Many of the minuscule objects aren’t clearly advertised.

[…]

But there is no doubt some online sellers deliberately trick customers into buying smaller and often cheaper-to-produce items, Witcher said. Common tactics include displaying products against a white background rather than in room sets or on models, or photographing items with a perspective that makes them appear bigger than they really are. Dimensions can be hidden deep in the product description, or not included at all.

In those instances, the duped consumer “may say, well, it’s only \$1, \$2, maybe \$3—what’s the harm?” Witcher said. When the item arrives the shopper may be confused, amused or frustrated, but unlikely to complain or demand a refund.

“When you aggregate that to these companies who are selling hundreds of thousands, maybe millions of these items over time, that adds up to a nice chunk of change,” Witcher said. “It’s finding a loophole in how society works and making money off of it.”

Defrauding a lot of people out of a small amount each can be a very successful way of making money.

### The Privacy Disaster of Modern Smart Cars

[2023.11.10] Article based on a Mozilla report.

### Ten Ways AI Will Change Democracy

[2023.11.13] Artificial intelligence will change so many aspects of society, largely in ways that we cannot conceive of yet. Democracy, and the systems of governance that surround it, will be no exception. In this short essay, I want to move beyond the “AI-generated disinformation” trope and speculate on some of the ways AI will change how democracy functions—in both large and small ways.

When I survey how artificial intelligence might upend different aspects of modern society, democracy included, I look at four different dimensions of change: speed, scale, scope, and sophistication. Look for places where changes in degree result in changes of kind. Those are where the societal upheavals will happen.

Some items on my list are still speculative, but none require science-fictional levels of technological advance. And we can see the first stages of many of them today. When reading about the successes and failures of AI systems, it’s important to differentiate between the fundamental limitations of AI as a technology, and the practical limitations of AI systems in the fall of 2023. Advances are happening quickly, and the impossible is becoming the routine. We don’t know how long this will continue, but my bet is on continued major technological advances in the coming years. Which means it’s going to be a wild ride.

So, here’s my list:

1. AI as educator. We are already seeing AI serving the role of teacher. It’s much more effective for a student to learn a topic from an interactive AI chatbot than from a textbook. This has applications for democracy. We can imagine chatbots teaching citizens about different issues, such as climate change or tax policy. We can imagine candidates deploying chatbots of themselves, allowing voters to directly engage with them on various issues. A more general chatbot could know the positions of all the candidates, and help voters decide which best represents their position. There are a lot of possibilities here.
2. AI as sense maker. There are many areas of society where accurate summarization is important. Today, when constituents write to their legislator, those letters get put into two piles—one for and another against—and someone compares the height of those piles. AI can do much better. It can provide a rich summary of the comments. It can help figure out which are unique and which are form letters. It can highlight unique perspectives. This same system can also work for comments to different government agencies on rulemaking processes—and on documents generated during the discovery process in lawsuits.
3. AI as moderator, mediator, and consensus builder. Imagine online conversations in which AIs serve the role of moderator. This could ensure that all voices are heard. It could block hateful—or even just off-topic—comments. It could highlight areas of agreement and disagreement. It could help the group reach a decision. This is nothing that a human moderator can’t do, but there aren’t enough human moderators to go around. AI can give this capability to every decision-making group. At the extreme, an AI could be an arbiter—a judge—weighing evidence and making a decision. These capabilities don’t exist yet, but they are not far off.
4. AI as lawmaker. We have already seen proposed legislation written by AI, albeit more as a stunt than anything else. But in the future AIs will help craft legislation, dealing with the complex ways laws interact with each other. More importantly, AIs will eventually be able to craft loopholes in legislation, ones potentially too complicated for people to easily notice. On the other side of that, AIs could be used to find loopholes in legislation—for both existing and pending laws. And more generally, AIs could be used to help develop policy positions.
5. AI as political strategist. Right now, you can ask your favorite chatbot questions about political strategy: what legislation would further your political goals, what positions to publicly take, what campaign slogans to use. The answers you get won’t be very good, but that’ll improve with time. In the future we should expect politicians to make use of this AI expertise: not to follow blindly, but as another source of ideas. And as AIs become more capable at using tools, they can automatically conduct polls and focus groups to test out political ideas. There are a lot of possibilities here. AIs could also engage in fundraising campaigns, directly soliciting contributions from people.
6. AI as lawyer. We don’t yet know which aspects of the legal profession can be done by AIs, but many routine tasks that are now handled by attorneys will soon be able to be completed by an AI. Early attempts at having AIs write legal briefs haven’t worked, but this will change as the systems get better at accuracy. Additionally, AIs can help people navigate government systems: filling out forms, applying for services, contesting bureaucratic actions. And future AIs will be much better at writing legalese, reducing the cost of legal counsel.
7. AI as cheap reasoning generator. More generally, AI chatbots are really good at generating persuasive arguments. Today, writing out a persuasive argument takes time and effort, and our systems reflect that. We can easily imagine AIs conducting lobbying campaigns, generating and submitting comments on legislation and rulemaking. This also has applications for the legal system. For example: if it is suddenly easy to file thousands of court cases, this will overwhelm the courts. Solutions for this are hard. We could increase the cost of filing a court case, but that becomes a burden on the poor. The only solution might be another AI working for the court, dealing with the deluge of AI-filed cases—which doesn’t sound like a great idea.
8. AI as law enforcer. Automated systems already act as law enforcement in some areas: speed trap cameras are an obvious example. AI can take this kind of thing much further, automatically identifying people who cheat on tax returns or when applying for government services. This has the obvious problem of false positives, which could be hard to contest if the courts believe that “the computer is always right.” Separately, future laws might be so complicated that only AIs are able to decide whether or not they are being broken. And, like breathalyzers, defendants might not be allowed to know how they work.
9. AI as propagandist. AIs can produce and distribute propaganda faster than humans can. This is an obvious risk, but we don’t know how effective any of it will be. It makes disinformation campaigns easier, which means that more people will take advantage of them. But people will be more inured against the risks. More importantly, AI’s ability to summarize and understand text can enable much more effective censorship.
10. AI as political proxy. Finally, we can imagine an AI voting on behalf of individuals. A voter could feed an AI their social, economic, and political preferences; or it can infer them by listening to them talk and watching their actions. And then it could be empowered to vote on their behalf, either for others who would represent them, or directly on ballot initiatives. On the one hand, this would greatly increase voter participation. On the other hand, it would further disengage people from the act of understanding politics and engaging in democracy.

When I teach AI policy at HKS, I stress the importance of separating the specific AI chatbot technologies in November of 2023 with AI’s technological possibilities in general. Some of the items on my list will soon be possible; others will remain fiction for many years. Similarly, our acceptance of these technologies will change. Items on that list that we would never accept today might feel routine in a few years. A judgeless courtroom seems crazy today, but so did a driverless car a few years ago. Don’t underestimate our ability to normalize new technologies. My bet is that we’re in for a wild ride.

This essay previously appeared on the Harvard Kennedy School Ash Center’s website.

### How .tk Became a TLD for Scammers

[2023.11.14] Sad story of Tokelau, and how its top-level domain “became the unwitting host to the dark underworld by providing a never-ending supply of domain names that could be weaponized against internet users. Scammers began using .tk websites to do everything from harvesting passwords and payment information to displaying pop-up ads or delivering malware.”

### Upcoming Speaking Engagements

[2023.11.14] This is a current list of where and when I am scheduled to speak: