Artificial intelligence is increasingly being used by governments to analyse information, support defence operations and review large volumes of grant applications. Supporters say it can reduce costs, improve efficiency and help public bodies process complex information at speed.
But recent developments in the technology sector have reignited debate about Ethical AI and the importance of keeping humans responsible for final decisions.
While AI systems can support analysis, experts warn they can also contain bias, make mistakes and produce misleading results. For that reason, many policymakers say the most important safeguard is maintaining human oversight.
Tech Industry Rethinks Military AI
The debate intensified following a dispute between the AI company Anthropic and the United States Department of Defense.
Anthropic launched legal action after the Pentagon labelled the company a “supply chain risk”. The conflict followed Anthropic’s refusal to remove safeguards that prevented its AI systems from being used for domestic mass surveillance or fully autonomous weapons.
The dispute highlights a wider shift across the technology sector.
In 2018, thousands of employees at Google protested against the company’s involvement in analysing military drone footage through a programme known as Project Maven. At the time, the company declined to renew the contract following internal pressure.
Since then, attitudes in Silicon Valley have changed significantly. Major AI developers including OpenAI, Google, Anthropic and others have signed defence contracts worth up to 200 million dollars each with the Pentagon to help integrate AI tools into national security systems.
Supporters argue that democratic governments need advanced AI capabilities to keep pace with global technological competition. Critics say these developments raise serious ethical questions about how the technology could be used.
AI Mistakes and Bias
Concerns are not limited to military applications.
A recent controversy involved officials working inside the US National Endowment for the Humanities who used ChatGPT as part of a process to review grant applications. The system was used to identify projects linked to diversity, equity and inclusion policies before officials decided which grants to cancel.
More than 1,400 grants were eventually terminated.
Critics say the episode highlights the risks of relying on automated systems without expert oversight. AI models can misinterpret information or reinforce bias present in their training data.
Because they generate answers based on patterns rather than true understanding, their conclusions can appear authoritative even when they are wrong.
Principles Behind Ethical AI
Governments and regulators have increasingly called for clearer standards to guide how artificial intelligence is used in public life.
Guidance published by GOV.UK outlines several core principles behind Ethical AI.
These include fairness, transparency, accountability and safety. Systems should also protect privacy, be explainable and operate in ways that benefit society.
Most importantly, humans must remain responsible for the outcomes produced by AI supported systems.
Keeping Humans in the Loop
Many researchers believe the safest model is a “human in the loop” approach.
In this system, AI can process large datasets, identify patterns and highlight potential issues. However, trained professionals review the results and make the final judgement.
This is particularly important in high risk areas such as defence, public funding and national security.
AI can analyse information faster than humans, but it cannot understand social context, ethics or the long term consequences of decisions in the same way people can.
Balancing Innovation and Responsibility
Governments around the world are exploring how AI could make public services more efficient. Automated tools could reduce administrative workloads and help institutions manage growing volumes of information.
But the recent controversies around military contracts and grant decisions show that technological innovation must be balanced with responsibility.
As AI becomes more deeply embedded in government systems, the debate around Ethical AI is likely to grow.
For many experts, the conclusion is clear. Artificial intelligence can support human decision making, but it should never replace it. Human oversight remains the most important safeguard when technology begins to shape decisions that affect people’s lives.
