There are numerous good places for promoting. A conversation with Claude is just not one in every of them.
Promoting drives competition, helps people discover latest products, and allows services like email and social media to be offered without spending a dime. We’ve run our own ad campaigns, and our AI models have, in turn, helped lots of our customers within the promoting industry.
But including ads in conversations with Claude could be incompatible with what we wish Claude to be: a genuinely helpful assistant for work and for deep pondering.
We wish Claude to act unambiguously in our users’ interests. So we’ve made a alternative: Claude will remain ad-free. Our users won’t see “sponsored” links adjoining to their conversations with Claude; nor will Claude’s responses be influenced by advertisers or include third-party product placements our users didn’t ask for.
The character of AI conversations
When people use engines like google or social media, they’ve come to expect a mix of organic and sponsored content. Filtering signal from noise is a component of the interaction.
Conversations with AI assistants are meaningfully different. The format is open-ended; users often share context and reveal greater than they’d in a search query. This openness is a component of what makes conversations with AI priceless, nevertheless it’s also what makes them liable to influence in ways in which other digital products aren’t.
Our evaluation of conversations with Claude (conducted in a way that keeps all data private and anonymous) shows that an appreciable portion involve topics which might be sensitive or deeply personal—the sorts of conversations you would possibly have with a trusted advisor. Many other uses involve complex software engineering tasks, deep work, or pondering through difficult problems. The looks of ads in these contexts would feel incongruous—and, in lots of cases, inappropriate.
We still have much to learn in regards to the impact of AI models on the individuals who use them. Early research suggests each advantages—like people finding support they couldn’t access elsewhere—and risks, including the potential for models to strengthen harmful beliefs in vulnerable users. Introducing promoting incentives at this stage would add one other level of complexity. Our understanding of how models translate the goals we set them into specific behaviors remains to be developing; an ad-based system could subsequently have unpredictable results.
Incentive structures
Being genuinely helpful is one in every of the core principles of Claude’s Structure, the document that describes our vision for Claude’s character and guides how we train the model. An advertising-based business model would introduce incentives that would work against this principle.
Consider a concrete example. A user mentions they’re having trouble sleeping. An assistant without promoting incentives would explore the varied potential causes—stress, environment, habits, and so forth—based on what is likely to be most insightful to the user. An ad-supported assistant has an extra consideration: whether the conversation presents a chance to make a transaction. These objectives may often align—but not at all times. And, unlike a listing of search results, ads that influence a model’s responses may make it difficult to inform whether a given suggestion comes with a business motive or not. Users shouldn’t should second-guess whether an AI is genuinely helping them or subtly steering the conversation towards something monetizable.
Even ads that don’t directly influence an AI model’s responses and as a substitute appear individually inside the chat window would compromise what we wish Claude to be: a transparent space to think and work. Such ads would also introduce an incentive to optimize for engagement—for the period of time people spend using Claude and the way often they return. These metrics aren’t necessarily aligned with being genuinely helpful. Essentially the most useful AI interaction is likely to be a brief one, or one which resolves the user’s request without prompting further conversation.
We recognize that not all promoting implementations are equivalent. More transparent or opt-in approaches—where users explicitly decide to see sponsored content—might avoid among the concerns outlined above. However the history of ad-supported products suggests that promoting incentives, once introduced, are inclined to expand over time as they grow to be integrated into revenue targets and product development, blurring boundaries that were over again clear-cut. We’ve chosen to not introduce these dynamics into Claude.
Our approach
Anthropic is concentrated on businesses, developers, and helping our users flourish. Our business model is easy: we generate revenue through enterprise contracts and paid subscriptions, and we reinvest that revenue into improving Claude for our users. It is a alternative with tradeoffs, and we respect that other AI corporations might reasonably reach different conclusions.
Expanding access to Claude is central to our public profit mission, and we wish to do it without selling our users’ attention or data to advertisers. To that end, we’ve brought AI tools and training to educators in over 60 countries, begun national AI education pilots with multiple governments, and made Claude available to nonprofits at a big discount. We proceed to take a position in our smaller models in order that our free offering stays on the frontier of intelligence, and we may consider lower-cost subscription tiers and regional pricing where there is evident demand for it. Should we’d like to revisit this approach, we’ll be transparent about our reasons for doing so.
Supporting commerce
AI will increasingly interact with commerce, and we stay up for supporting this in ways in which help our users. We’re particularly interested by the potential of agentic commerce, where Claude acts on a user’s behalf to handle a purchase order or booking end to finish. And we’ll proceed to construct features that enable our users to search out, compare, or buy products, connect with businesses, and more—once they decide to achieve this.
We’re also exploring more ways to make Claude a focused space to be at your best. Users can already connect third-party tools they use for work—like Figma, Asana, and Canva—and interact with them directly inside Claude. We expect to introduce many more useful integrations and expand this toolkit over time.
All third-party interactions will likely be grounded in the identical overarching design principle: they ought to be initiated by the user (where the AI is working for them) quite than an advertiser (where the AI is working, not less than partly, for another person). Today, whether someone asks Claude to research trainers, compare mortgage rates, or recommend a restaurant for a special day, Claude’s only incentive is to present a helpful answer. We’d wish to preserve that.
A trusted tool for thought
We wish our users to trust Claude to assist them keep pondering—about their work, their challenges, and their ideas.
Our experience of using the web has made it easy to assume that promoting on the products we use is inevitable. But open a notebook, pick up a well-crafted tool, or stand in front of a clean chalkboard, and there aren’t any ads in sight.
We expect Claude should work the identical way.
