AI coding tools accelerate retreat from open-source licensing as developers struggle to protect their work
From Bear's licensing change to HashiCorp's Business Source License, creators face new pressures at a time when artificial intelligence makes code cloning faster than ever
Creating a competitor to someone's life's work now takes about as long as ordering coffee.
Herman, creator of the minimalist blogging platform Bear (used to host Nearly Right), discovered this when people began forking his code and launching rival services within hours. His response was swift and painful, abandoning the MIT licence that had welcomed collaboration for years, switching instead to the restrictive Elastic Licence.
"It hurts to see something you've worked so hard on for so long get copied and distributed with only a few hours of modification," Herman wrote on his blog. "It hurts to have poured so much love into a piece of software to see it turned against you and threaten your livelihood."
This isn't just one developer's bad day. Herman's anguish signals a seismic shift reshaping software development, where artificial intelligence has weaponised the very openness that built the modern internet. The collaborative spirit of open source—once protected by the sheer difficulty of meaningful copying—now faces an existential threat from AI tools that can clone, modify, and deploy competing projects faster than their creators can respond.
The evidence is stark. GitClear's analysis of 211 million lines of code from major technology companies reveals AI coding assistants correlate with a fourfold increase in code cloning. What Herman describes—typing "Create a fork of this repo and change its name to something cool and deploy it on an EC2 instance"—isn't hyperbole. It's the new reality.
When collaboration becomes competition
The timing couldn't be worse. This AI-powered cloning explosion arrives precisely when open-source creators were already fighting for financial survival. The traditional model—give away the software, monetise through services or enterprise features—worked brilliantly when building meaningful competition required months of dedicated expertise.
Those days are dead.
GitHub Copilot, now used by millions of developers, can generate entire functions from simple comments. Cursor AI explicitly markets itself as an AI-powered development environment designed for rapid code modification. The barrier between inspiration and implementation has evaporated.
Consider the cruel mathematics, a developer spends years perfecting their creation, releases it openly to benefit the community, then watches AI-assisted competitors launch rival services before their original announcement gains traction on social media. Herman's experience with Bear isn't unique—it's becoming routine.
The casualty list grows longer each month. Major technology companies, built on open-source foundations, now abandon the very licences that enabled their success.
The great licensing retreat begins
HashiCorp fired the starting gun in August 2023. The company behind Terraform—infrastructure-as-code software used by millions of developers—abandoned its Mozilla Public Licence for the Business Source Licence. Their reasoning was blunt, vendors were "taking advantage of pure OSS models and the community work on OSS projects for their own commercial goals, without providing material contributions back."
The development community erupted in outrage, predicting user exodus and project abandonment. Instead, HashiCorp's usage continued climbing, validating a painful truth, protective licensing works.
This followed a template pioneered by MongoDB, which triggered similar fury in 2018 with its licensing change—then watched usage increase sixfold. The pattern was becoming clear, temporary community backlash, followed by sustained growth under protective licensing.
Elastic provides the most instructive case study. When the search engine company switched from Apache 2.0 to restrictive dual licensing in 2021, it specifically targeted Amazon Web Services' practice of offering competing managed services without contributing back. AWS responded by forking Elasticsearch entirely, creating the rival OpenSearch project.
Rather than splitting the market, the move clarified it. By 2024, Elastic felt secure enough to add open-source AGPL licensing as a third option—essentially declaring victory. The company had forced competitors to build their own platforms while maintaining control over their original creation.
The architect's diagnosis
Bruce Perens has watched this crisis build for decades. The co-founder of the Open Source Initiative and creator of the original Open Source Definition calls the current situation a systematic failure, not corporate greed.
"Our licences aren't working anymore," Perens argues. "We've had enough time that businesses have found all of the loopholes and thus we need to do something new."
His diagnosis cuts deeper than individual company decisions. Open source has become "a great corporate welfare program" where "our users are the richest companies in the world" whilst developers "probably go un-compensated." This corporate-developer power differential has reached breaking point as AI tools make exploitation effortless.
The scale of dysfunction is staggering. The Open Source Initiative now recognises over 100 different software licences, creating what Perens calls "a tremendous non-compliance problem." One executive told him their compliance department costs $7 million annually—just to navigate the legal maze of sharing software freely.
Amanda Brock, CEO of OpenUK and a board member of the Open Source Initiative, acknowledges the crisis whilst defending collaborative principles. "The friction of requiring commercial payment, patent licences etc—which cannot happen with a true open source licence—interrupts that free flow and jeopardises the ability to use and reuse code," she explains.
Yet even Brock welcomed Elastic's partial return to open source, recognising that pure ideological positions may be unsustainable in the AI era.
The search for middle ground
Innovation emerges from crisis. Rather than abandoning collaborative principles entirely, developers are inventing hybrid approaches that preserve sharing whilst protecting creators.
The Business Source Licence, pioneered by MariaDB in 2016 and adopted by HashiCorp, CockroachDB, and Sentry, offers an elegant compromise, free use for everyone except direct competitors, with automatic conversion to traditional open source after four years. It's collaborative development with training wheels—protection that expires once projects mature.
Sentry recently introduced "fair source" licensing, combining transparency with time-limited restrictions. These experiments seek that elusive balance between openness and sustainability that traditional licensing failed to achieve.
The trend extends beyond individual solutions. FOSSA's 2024 industry analysis found source-available licences gaining momentum across technology companies. Even controversial cases like PearAI—which forked an AI coding assistant and switched from Apache to proprietary licensing—reflect growing recognition that unlimited openness may be unsustainable.
The irony runs deep, AI tools that threaten open source were largely trained on freely shared code. GitHub Copilot learned from millions of publicly available repositories, many using the permissive licences now under threat. The technology that democratised coding may ultimately force its restriction.
What we lose and what we gain
The stakes extend far beyond software licensing. Open source built the internet, powers smartphones, and runs the cloud infrastructure supporting global commerce. Linux alone demonstrates how collaborative development can create technology supporting billions of people.
Yet the evidence suggests protective licensing doesn't kill innovation—it reshapes it. MongoDB's controversial licensing change preceded years of accelerated development and market expansion. HashiCorp's tools remain essential infrastructure despite business licence restrictions. Elastic's protective approach enabled both community development and commercial growth.
The question isn't whether restrictions reduce innovation, but whether they redirect it. Traditional open source assumed abundance—unlimited developer time, infinite collaborative goodwill, and benevolent corporate participation. AI-accelerated competition has revealed these assumptions as dangerously naive.
New licensing models may actually strengthen collaboration by ensuring it remains sustainable. If creators abandon open development because exploitation becomes too easy, everyone loses. Better to share code with reasonable restrictions than not share it at all.
The future of sharing
The transformation resembles other technological disruptions. Photography forced painters to discover new artistic purposes. The printing press made scribes obsolete but democratised knowledge. AI coding assistants are forcing developers to reimagine the relationship between creation and sharing.
Herman's decision to restrict Bear's licence represents one response, accepting reduced collaboration for greater control. But the industry is developing more sophisticated solutions that preserve sharing whilst addressing legitimate exploitation concerns.
Bruce Perens works on what he calls "post-open source" licensing, seeking frameworks that ensure fair compensation for creators whilst maintaining collaborative development. Companies experiment with fair source and business source approaches, gradually defining new standards for sustainable sharing.
The question isn't whether open source will survive AI acceleration—it's how it will evolve. Early evidence suggests the community is adapting rather than retreating, developing licensing innovations as sophisticated as the technology that threatened traditional models.
Herman's anguish was real and justified. Watching years of careful work copied in hours would devastate anyone. But his experience also reflects the growing pains of an ecosystem discovering how to remain both collaborative and sustainable in an age of artificial intelligence.
The future of software development hangs on resolving this tension. The collaborative spirit that built the modern internet must find new expressions that honour both the sharing impulse that drives innovation and the legitimate interests of those who create the code powering our digital world.
Getting this balance right may determine whether AI amplifies human creativity or simply automates its exploitation.