For ten days in May 1995, Brendan Eich occupied one of those peculiar rooms in technological history where a deadline, a handful of borrowed ideas, and an unfinished product all arrive at once. Netscape needed a small programming language for its coming browser release. Eich had joined expecting to put Scheme in the browser. Management wanted something that looked more like Java. The result, code-named Mocha, carried Scheme’s first-class functions, Self’s prototype model, and a Java-like surface. At the end of the sprint, he demonstrated it to Netscape’s engineering staff.
The demo worked. That sentence contains the beginning of a language and the beginning of a maintenance problem. Mocha became LiveScript, then JavaScript in a December announcement with Sun Microsystems. The name implied kinship with Java that the designs did not share. Netscape shipped it, Microsoft implemented its own version, and the web acquired a way to do more than display documents. Buttons could respond. Forms could reason. Pages could change without asking the server to redraw everything.
“Ten days” became the neat version of the origin story, compact enough for a conference introduction. The less neat version matters more. The prototype had to be integrated into Netscape Navigator, exposed to web pages, debugged across platforms, documented, standardized, and extended without breaking what people had already written. Eich also designed SpiderMonkey, the first engine to execute the new language. A quick draft turned into a compatibility promise that outlived Netscape itself.
A language escapes its deadline
Eich was not a novice improvising from nothing. He had studied mathematics and computer science at Santa Clara University, earned a master’s in computer science at the University of Illinois Urbana-Champaign, and spent seven years at Silicon Graphics working on operating-system and network code. The ten-day sprint compressed experience. It did not replace it.
It also produced scars. Eich acknowledged bugs early, and the language’s compact first design accumulated behaviors that could not simply be removed once sites depended on them. JavaScript’s success made taste subordinate to compatibility. Standards bodies, browser engineers, and developers had to improve the language while keeping yesterday’s pages alive. The original author became one participant in a much larger process.
Netscape
Make the page programmable. Mocha becomes JavaScript.
Mozilla
Give open browser code an institution and a community.
Brave
Make privacy and user control part of the browser default.
This is one of the useful tensions in Eich’s career. He is remembered for a compressed act of creation, yet much of his work has involved systems that resist individual control. A programming language gains committees. Open-source code gains maintainers. A browser gains users whose old extensions, favorite sites, and ingrained habits limit every clean-sheet redesign.
Only the Web is so big, with such reach to billions of users, that no one owns it. This means it will always be contested ground.Brendan Eich, 2016
The institution after the company
In 1998, Netscape released browser source code and Eich helped found the Mozilla project with Jamie Zawinski and others. Mozilla was initially a place to coordinate contributions. It became something rarer: an attempt to preserve browser competition after the company that opened the code lost its independence. AOL acquired Netscape in 1999 and closed its browser unit in 2003. Eich and Mitchell Baker helped spin out the nonprofit Mozilla Foundation.
Eich served as Mozilla’s chief architect, then became chief technology officer of the Mozilla Corporation in 2005. Firefox gave the project a product ordinary users could choose, while Mozilla gave contributors and engineers an organization able to hold the code, employ people, and negotiate with larger companies. JavaScript had made the browser a runtime. Mozilla treated the browser as public infrastructure.
The Mozilla chapter ended abruptly. Eich was appointed CEO on March 24, 2014. The appointment triggered protests and boycotts after renewed attention to his 2008 donation supporting California Proposition 8, which sought to ban same-sex marriage. On April 3, he voluntarily resigned. Mozilla said its board had invited him to remain in another senior role, which he declined. Eich wrote that the mission was bigger than any individual and that, under the circumstances, he could not be an effective leader.
A former Mozilla colleague remembered a San Francisco outing with Eich and Andreas Gal that turned into a two-hour tour of the wrong institutions, all with some combination of “Academy” and “Science” in their names. The three locals arrived late to a destination that was not far from where they began. Infrastructure expertise, apparently, does not guarantee map expertise.
Colleagues describe Eich as knowledgeable, strongly opinionated, and ready to move from low-level implementation details to economics or web governance. His public interviews have the same density. A question about advertising can produce a tour through auctions, tokens, browser state, fraud, and cryptography. The through line is control: which party can see the data, set the default, or close the platform?
The browser as a boundary
In 2015, Eich and former Mozilla engineer Brian Bondy co-founded Brave. Its browser began with a plain proposition: block trackers and intrusive ads by default. The architecture eventually moved onto Chromium, the open-source base also used by Chrome and other browsers. Brave’s differentiation would live in what it removed, what it added, and which interests its defaults served.
Blocking was only half the argument. The web still had publishers to fund, creators to support, and advertisers willing to pay for attention. Brave experimented with privacy-preserving ads selected without remote behavioral profiles, a rewards system built around the Basic Attention Token, creator contributions, subscriptions, a self-custody wallet, and paid services. The combination is deliberately untidy. It tries to alter the economics around surveillance without pretending that content arrives for free.
The product widened. Brave Search launched in 2021 with an independent index. Brave Wallet put multi-chain asset management inside the browser. Leo added an AI assistant in 2023. Ask Brave, released in 2025, combined conventional search results with conversational answers. By late 2025, the company said its browser had crossed 100 million monthly active users.
Those additions make the privacy question harder, not easier. A browser assistant can read a page to summarize it. An agent can take actions. A search system can use queries to generate answers. Each new ability increases the value of context while increasing the cost of mishandling it. In November 2025, Eich co-authored a Brave post describing an experimental version of Leo using trusted execution environments, designed to let users verify both the model and the privacy of the conversation.
The continuity with JavaScript is easy to miss. Browser code has always required a theory of authority: which page may open a window, read a cookie, call a network address, or touch another page’s data. Decades of security engineering built boundaries around those actions. An agent adds ambiguity because its input is ordinary language gathered from places with conflicting motives. A travel page may contain a useful itinerary, an advertisement, a user review, and a concealed instruction aimed at the assistant. All are text. Only some deserve authority. Solving that problem means more than attaching a chatbot to a tab. It requires permissions, isolation, visible confirmation, and a reliable distinction between content the agent may summarize and commands it may obey. For Eich, this is familiar territory in a new register: powerful behavior inside the browser, surrounded by adversarial input and compatibility pressure.
By March 2026, his public focus had moved to browser agents and indirect prompt injection: hostile instructions hidden in content that an agent reads. The problem is almost a mirror image of 1995. Then, the breakthrough was letting a page contain code that could act. Now the task is deciding what an AI agent should trust when a page contains language that looks like an instruction.
Be a human, write and read code.Brendan Eich at Permissionless IV, 2025
Defaults are where values become behavior
Eich’s career is often told as a sequence of nouns: JavaScript, Mozilla, Brave. The verbs are more revealing. Prototype. Release. Standardize. Spin out. Block. Index. Verify. Each one moves an idea from a statement into a mechanism that other people have to live with.
The practical lesson for builders is not to romanticize speed. A ten-day prototype is impressive because decades of people made it useful afterward. The lesson is to notice where a default becomes infrastructure. JavaScript shipped with a browser and became available to every page. Mozilla put open code inside an organization able to sustain it. Brave puts privacy choices in the path of everyday browsing instead of leaving them as optional homework.
There is no final browser. The job keeps expanding as the web absorbs payments, media, identity, search, and AI. Every expansion creates another argument about who gets to observe, decide, and profit. Eich’s second draft is not a replacement for the first. It is an attempt to redraw the boundary around it, one release at a time.