Bill Gates: AI Could Trigger Mass Unemployment, Spread Cyber Attacks, and Increase Bio-Terrorism Risks
Bill Gates warns in a recent lengthy article that if countries do not establish governance and redistribution mechanisms in advance, AI could simultaneously trigger mass unemployment, spread cyber attacks, and increase the risk of bio-terrorism.
Gates states that AI will impact both white-collar and blue-collar jobs in fields such as law, customer service, healthcare, software, and manufacturing, and the speed of this change could be measured in decades rather than generations as seen in past technological revolutions. He particularly notes that entry-level and mid-level positions face greater replacement pressure, and the skills required for newly created jobs may not be quickly acquired by those affected.
He suggests that if human society continues to rely on work for basic living security, policies must be designed before unemployment and low employment arise, including shortening work hours, protecting certain "human-only" professions, and exploring taxes on AI computing resources or robots to support public services and income security.
Regarding cybersecurity, Gates believes AI is lowering the technical barriers for attackers to discover software vulnerabilities, create fraudulent content, launch phishing attacks, and organize large-scale cyber intrusions. Critical infrastructure such as hospitals, financial institutions, water supply systems, power grids, and government welfare management systems could become high-value targets, with the speed of attack capability enhancement potentially outpacing defensive repairs.
On biosecurity, he warns that the same models that assist in drug, vaccine, and disease research could also lower the knowledge and resource barriers for designing dangerous pathogens. Gates does not claim that bio-terror attacks have already occurred, but believes AI will empower malicious actors with previously limited capabilities, and defensive and offensive biological knowledge is difficult to separate completely.
He advocates for establishing new domestic and international systems to coordinate AI governance, arguing that existing institutions are not designed for a rapidly spreading, cross-sector technology that affects economy, security, and social life. He states that if a credible, globally coordinated AI deceleration plan exists, he might support it, but current geopolitical and economic incentives are driving all parties to accelerate development.
In market mechanisms, AI investment is flowing into models, chips, data centers, and application automation, with companies improving efficiency by replacing some manual processes; at the same time, there will be new demands for cybersecurity, model evaluation, biosecurity, identity verification, critical infrastructure protection, and retraining services. Repetitive knowledge work and entry-level positions lacking bargaining power are under pressure, while positions with proprietary data, regulatory licenses, physical execution capabilities, and high responsibility attributes are harder to fully replace in the short term.
Source: Public Information
ABAB AI Insight
Gates is not suddenly turning to AI pessimism. Microsoft initially pushed personal computers into offices and homes, and the Gates Foundation later engaged in vaccine, infectious disease, and global health work; thus, he sees both the potential benefits of AI in drug development, diagnostics, and education, as well as the dual-use risks after the diffusion of biological knowledge. His emphasis that "many jobs will disappear forever" positions AI as a structural change in the labor market rather than a simple productivity upgrade.
In terms of capital pathways, AI benefits will first concentrate in companies that possess computing power, data, models, cloud platforms, and corporate distribution channels; costs may be transferred to workers and governments through layoffs, wage pressures, training burdens, and public finance. The proposed tax on computing resources or robots by Gates essentially attempts to redirect a portion of automation gains from the capital side back to social security, education, and public services to avoid a disconnect between productivity growth and residents' income.
Historically, the Industrial Revolution increased long-term productivity but also initially led to deteriorating factory labor conditions, skill obsolescence, and income disparity, ultimately giving rise to labor unions, compulsory education, social insurance, and labor regulations. The internet first compressed retail, media, and customer service sectors, then created platform economies and new technology jobs. The difference with AI is that it can simultaneously affect language, code, images, decision support, and physical robot control, so the impact may no longer be gradual across single industries but could occur synchronously across occupational levels.
Essentially, this is about regulatory change. The reason AI risks require new institutions is not that models inherently lead to disasters, but because the market will prioritize faster deployment, lower costs, and stronger capabilities without automatically bearing sufficient costs for retraining the unemployed, ensuring critical infrastructure safety, or addressing biological risks. Only by incorporating safety testing, accountability, capability access, employment buffers, and cross-border coordination into the rules can the alignment of corporate profit maximization and social risk minimization be achieved.
ABAB News · Cognitive Laws
Efficiency belongs to capital, while risks are often borne by society.
The faster the technology diffusion, the more expensive the institutional patches.
Capabilities are neither good nor evil; authority determines the boundaries of harm.