An opinion piece argues that public fears about artificial intelligence are being used to restrict open-source AI models, a move the writer says would concentrate power in the hands of a few large technology companies. The writer says AI is already causing harm to workers, the environment and communications, and that these problems are getting less attention than apocalyptic warnings.
The writer lists several existing harms. Companies have started laying off workers whose jobs AI models can now perform, and data centres have driven up energy consumption and carbon footprints. Malicious actors are using AI agents to automate disinformation, algorithms are showing bias, and AI is being used to identify targets in warfare, which the writer says has led to greater loss of human life. These problems, the writer says, demand serious regulatory attention, yet current policy debates are rarely grounded in them.
Doomsday warnings draw attention from immediate harms
The writer says global discussion has been dominated by apocalyptic predictions, promoted by some within the AI industry and echoed at the highest levels of decision-making. The examples cited include Senator Bernie Sanders, who has championed bans on superintelligence, cautionary messages from the Vatican, and former Secretary of State Hillary Clinton, who called open-source AI a real threat.
The writer says it is reasonable for global leaders to take such industry-amplified warnings seriously. However, the focus on “existential risk” has inadvertently pulled attention away from the crises AI is already causing, the writer argues.
How open-source limits could entrench corporate control
The writer argues that public panic is being weaponised to build a modern regime of digital colonisation, which the writer describes as the concentration of power in a few AI corporations in a handful of countries. Open-source models, which are free or cheaper to use and can be modified to fit users’ needs, would be the target of regulations justified by the doomsday narrative. The logic presented to lawmakers, the writer says, is that if advanced AI is a potential weapon of mass destruction, only a trusted few should develop it.
Under the banner of safety, the writer says, tech giants are trying to write rules that protect their monopolies by shutting out smaller, cheaper competitors. The writer compares this to historical colonial tactics that stopped local industry from developing so that colonised regions would depend on the empire’s exports.
Why developing nations need open-source AI
The writer says the world currently depends on frontier AI systems built by tech giants, and that the Global South lacks the resources to build its own models from scratch. The writer puts the cost of that at upwards of $100m in compute power alone. Dominant models are also costly and often fail in local contexts, the writer argues.
As an example, the writer points to Kenya’s recent rollout of the AI-driven Social Health Authority system. In the writer’s account, the imported model was built on assumptions that did not accurately capture the reality of Kenya’s vast informal economy. As a result, the system systematically overestimated incomes and overcharged the poorest citizens for healthcare premiums.
The writer argues that frontier systems must be adapted to local data, and that open-source AI is the only viable bridge for local innovators. Developing nations, the writer says, need AI that can help with energy, healthcare, agriculture, manufacturing and climate mitigation rather than conversational chatbots. The writer also warns that monopolies lock users into cycles of data extraction and cognitive dependency, pointing to early signs of cognitive degradation as people hand over critical thinking, creativity and agency. Wealthy nations in the Global North face the same threat, the writer adds.
The writer’s proposed path for AI safety
The writer rejects calls for AI bans. Abandoning AI, the writer says, would mean giving up a technology that could speed up the discovery of new drugs and solutions to energy and climate crises, with AlphaFold and MOFSeq cited as evidence. Banning open-source models is also a poor choice, the writer argues, because safety does not negate equitable access to AI.
The writer calls for funding robust safety research and rigorous testing for near-term breaches, without letting fears of hypothetical futures justify consolidating power in the present. Safeguarding AI, according to the writer, means keeping its development decentralised, transparent and globally distributed. It also means defending open-source innovation as a public good, addressing the human cognitive toll, and keeping accountability anchored in humans rather than in AI agents.
