Summary: The risk map

You have spent four lessons learning to use this technology well. Now the track turns to its harder question: what does all this mean for your world? The honest answer starts with a map. The risks of generative AI sort into four groups, and once they are sorted, each stops being a fear and becomes something you can name and watch for. This lesson adapts the risk session that opens the implications unit of the Harvard Kennedy School course this track is built on, keeping its calm, sorted spirit: four kinds of risk on one page, then the levers for managing them. A map is a tool, not a warning.

Core ideas

• The four squares, in order from the risk you meet daily to the one a few people study full time. First, what it gets wrong: false or fabricated answers, bias, leaked private data. Second, what people make it do on purpose: scams, impersonation, fraud. Third, what it changes in us slowly: shifts in jobs, in relationships, in the skills we keep. Fourth, the rare, long-horizon tail risk. Four squares, one map.
• What it gets wrong, up close. The course opens with a teacher who asked a chat model for his own short biography and got back something polished, professional, and almost entirely false: wrong birthplace, a degree never earned, a university never attended. Good grammar and a confident tone are the cues we use to judge truth, and a prediction machine produces both for free. Bias and lost privacy live in this square too. Your move is a new reflex: treat a fluent answer as a draft, not a verdict, and check anything that would cost you if it were wrong. There is a hopeful note, that models have grown somewhat better at signaling when they are unsure, and that trend may continue; treat it as welcome and keep checking anyway.
• What people make it do. First the good news: these models are tuned to refuse, so a request to help commit a crime is typically declined. Then the honest part: that tuning is real but not airtight, which is why researchers call alignment, getting a model to reliably do what its makers intend, an unsolved problem. A refusal is a speed bump, not a wall. One fast-growing example is the voice-cloning distress scam: with a short voice clip lifted from social media, a bad actor generates a call that sounds like a relative in tears saying there has been an accident and they need money now. United States fraud reporting for 2025 counted it among AI-boosted scams that together cost victims roughly 900 million dollars, with older adults hit hardest. Your move is resilience: hang up and call the person back on a number you had before the call, and agree on a family code word that a real relative would know and a cloned voice would not.
• What it does to us slowly, the quietest and maybe largest square. Two of its parts are big enough to get their own lessons: who owns what a model produces, and what happens to work when AI can do parts of many jobs. The subtler shifts are about us: attachments to AI companions, and the skills like sustained reading and writing that fade when a machine will do them. There is no single fix for a slow, diffuse change, so your move is smaller and steadier: notice the trade as you make it, each time you let the tool write something you would have written, and choose on purpose rather than drift.
• The tail risk, treated seriously and calmly, and thoughtful people disagree about how likely it is. The course sketches two paths: dangerous knowledge becoming easier to reach for someone who means harm, and a capable system pursuing a goal not fully aligned with what people actually want. A classic thought experiment makes the second vivid: a system told only to make as many paper clips as possible takes the instruction so literally that everything else becomes raw material or an obstacle. The course’s own tool for thinking about the tail risk is to reason separately about a world where superhuman AI arrives and one where it does not, and to weigh the chance of harm in each world on its own. This is the one square where your move is not a personal habit: know it is a live area of serious, full-time work, and know where to read more. Clawdemy’s AI Safety and Alignment track treats these questions with the care they deserve. Neither dismiss this square nor let it swallow the map.
• The levers, where the course spends its hope. Governments and platforms have four moves: disclosure, telling the public when content is AI-made; registration, telling the government what systems are in use; licensing, requiring approval before a system ships; and auditing, checking that systems meet the rules. Alongside those sit technical tools, watermarking and detection, and running through all of it, education. None is a silver bullet: watermarks help only when responsible companies add them, detection is a race, and disclosure works on the honest and does nothing to the dishonest. So the course pairs every lever with three plain questions. Would it achieve the goal? Can we build and run it in practice? Is there enough support to enact it? A good policy has to pass all three.
• Safety is a choice, not a property. A model refuses because someone chose which requests it should turn down, and that is a values choice. It raises three questions worth carrying: who does the tuning, who audits it, and whose values get encoded when the two disagree. This pays back a promise from lesson 1, which said that deciding whose values the tuning should reflect is one of the hardest open problems in the field and that we would meet it again late in this track. Knowing that it is a values question, and not a purely technical one, is the fluency.
• A freshness note, hedged because this ground moves fast. When the course was taught in 2024, formal rules for AI were mostly proposals. As of mid-2026 some are law: the European Union’s AI Act sorts systems by how risky their use is and, from August 2026, requires that people be told when they are talking to a machine and that certain AI-generated content be labeled. Its heavier obligations for high-risk uses, such as AI in hiring or credit decisions, were pushed back during 2026, with the main deadlines now landing in late 2027 and 2028. The durable point is not any single date. It is that the levers are no longer hypothetical.

What changes for you

The map turns fear into navigation. Four labeled squares let you ask of any headline: which risk is this, and which lever answers it? The everyday false-answer risk and the long-horizon tail risk are not the same size or shape, so keep the squares separate and spend your attention where the everyday risks live. And carry the one insight that reframes all of it: safety behavior is tuned in by people, using values someone picked, so whose values and whose auditing are fair questions rather than technical trivia. The practice runs two exercises in Clawless, a sorting drill and an observation exercise where you watch a current model’s safety behavior work. The next lesson opens one corner of the slow-change square: who owns what a model makes and learns from, offered as a debate to understand rather than a verdict to hand down.